Вы находитесь на странице: 1из 3

Journal of Comparative and Physiological Psychology 1965, Vol. 59, No.

2, 298-300

ACQUISITION OF SIDMAN AVOIDANCE RESPONDING AS A FUNCTION OF S-S INTERVAL


RUSSELL C. LEAF1 Squibb Institute for Medical Research Procedures for producing rapid and reliable lever-press Sidman avoidance acquisition in 100% of an unselected group of rats are described. Groups of 12 rats were exposed to schedules with an R-S interval of 20 sec., at S-S intervals of 1, 3, 5, 10, and 20 sec., with a constant power shocker. 1st session response levels were a monotonically decreasing .function of S-S interval. The hypothesis that acquisition probability is due to reduction of shock frequency was supported, but acquisition levels of responses with spaced IRTs were unrelated to S-S intervals.

Each S was given one 6-hr, acquisition session. The naive S was placed in a darkened experimental chamber. After 10 min., a 20-v. General Electric 304 house light was turned on, and the experimental contingencies started. The responseshock (R-S) interval was 20 sec. and shock duration 0.3 sec. for all Ss. Lever presses did not alter shock duration. A small pilot light mounted above METHOD each lever was illuminated for 40 msec, each time the lever was depressed. Avoidance contingencies Subjects were programed on only one lever for each S. The The Ss were 40, 250-300 gm, male and 20, 200- right lever was ,the avoidance lever for half the 250 gm. female albino rats purchased from Camm male and half the female Ss, and the left lever Research Institute, Wayne, New Jersey. was the avoidance lever for the remaining /Ss. The Ss were ran in 12 replications of 5 Ss. Apparatus Each replication contained only male, or only The experimental space was a modified Lehigh female, Ss. Within a replication, Ss were randomly Valley Electronics No. 1316 chamber, 7% X 12 X . assigned to order of training and to one of five 8 in. It had two LVE 1352 levers mounted sym- S-S intervals, 1, 3, 5, 10, and 20 sec. metrically on one 8 in. wall. The centers of the Stanley A. Muller provided assistance in analysis of data, and Peter L. Carlton commented on preliminary manuscript drafts of the work described.
1

Although many experiments have replicated Sidman's (1953) finding that rats can be trained to perform avoidance responses without exteroceptive warning stimuli, little parametric information on Sidman avoidance acquisition has been reported. This lack is probably due to the fact that methods for producing reliable lever-pressing Sidman avoidance learning in all Ss, from an unselected group of rats, are not readily available. The Es who have trained large groups of & have sometimes CVVeissman, 1962) reported a large percentage of training failures, even when the behavior was: studied for many training sessions. The present paper reports procedures for producing reliable and rapid acquisition of Sidman avoidance responding in a single experimental session. The shock-shock (S-S) interval was systematically varied in order to obtain data on the role of shock density reduction in Sidman avoidance acquisition. The only unusual feature of the training conditions was the use of a relatively constant power source (Campbell & Teghtsoonian, 1958; Hill, Flanary, Kornetsky, & Wikler, 1962), rather than constant current, shock source.

levers were 1V4 in. above the grid floor of the chamber, and were 5 in. apart. The chamber, in an LVE 1316 C sound insulated cubicle, was placed in a room separate from automatic programing and recording equipment. The chamber blower fan and a masking white noise served to minimize the effects of extraneous stimuli. The shock source was a modified LVE 1311 shocker. It consisted of (a) a step-up isolation transformer adjusted so that the voltage in its secondary coil was 725-v. ac, (t>) a 100 K ohm shunt across this secondary, (c) a 136 K ohm resistance in series with it, (d) an oil-immersed grid scrambler, (e) 40-ft. parallel wire-grid cable, (/) grid, and (g) S. This arrangement produced about 60 v. maximum across S, measured on a cathoderay oscilloscope (Tektronix Model 555), at relatively constant power, after attenuation by the capacitative reactance of scrambler, cables, and grid. Procedure

RESULTS

All /Ss in the 1- and 3-sec. S-S interval groups and 11 /Ss in the 5-sec. group learned to avoid reliably, but only 4 /Ss in the 10-sec. and 1 S

298

SUPPLEMENTARY REPORTS

299

Range

\Interquartile J Ranee

Mean

AVOIDANCE LEVER TOTAL SHOCKS

|" 3" 5" 10" S-S Interval

20

3" 5"

10" S-S Interval

FIG. 1. Means, interquartile ranges, and ranges of relevant lever responses, and total shocks for experimental groups with acquisition S-S intervals of 1, 3, 5, 10, and 20 sec. in the 20-sec. group did so. The reliable Ss all made at least 1,000 responses on the avoidance lever, spent less than 50% of the time in S-S intervals, and showed steady rates of avoidance responding during the last hour of the session, as judged from cumulative records. None of the & that failed to learn reliably had steady terminal rates, and all of these Ss spent more than 50% of the session in the S-S intervals and made fewer than 1,000 responses on the avoidance lever. The effect of the S-S interval on avoidance lever responding is shown in Figure 1. Avoidance responding was a significant (H 28.1, df = 4, p < .001) monotonic decreasing function of S-S interval. As Figure 2 shows, however, this monotonic function was due primarily to the levels of (a) rapid burst-like responses with interresponse times (IRTs) less than 2 see., and (6) responses that occurred after the receipt of an R-S shock (IRTs > 20 sec.). High levels of spaced responses, with IRTs of 2-20 sec., were observed; but these levels were not monotonically related to S-S interval. Each of the nine class intervals of 2 sec. for IRTs from 2-20 sec. showed an inverted U-shaped function relating response level to S-S interval, with peaks at either 3 or 5 sec. All nine functions were significant (H > 18.5, df = 4, p < .001 for each distribution). In order to assess the effect of S-S interval, additional analysis was carried out on the data of only those <Ss that learned reliably (1,000 responses, 50% S-S time, steady terminal rates) and were trained with 1, 3, and 5 sec. S-S intervals. These & showed no significant relationship between avoidance responding and S-S interval (H = 0.6, df - 2, p > .50). When avoidance responses were categorized by IRTs, as shown in Figure 2, 0-2 sec. IRT responses were a nonsignificant decreasing function of S-S interval (H = 2.0, df - 2, p > .20), 2-20
RELEVANT LEVER IRT DISTRIBUTIONS

2142 1930 1700 1600 1500 1400 1300

r 1

o n

' " 3 " 5" 10" 20 Means

I
O '( 3

Interquartile Range Range

U> 1200 1100 g 1000 2 900 K 800


700 600 500 400 300 200 100

o ct
[]

''

A
[ ^
H1 []

1
IT 5" 100'-2"IRTS

1
1"3"5" 10" 2"-20"IRTS.

11
Iff 1'3" 510" >20" IRTS

1
2

20-

FIG. 2. Interresponse time distributions for Ss with acquisition S-S intervals of 1, 3, 5, 10, and 20 sec. (Means, interquartile ranges, and ranges of the frequency of responses with IRTs of 0-2 sec., 2-20 sec., and greater than 20 see. are shown for each S-S interval.)

300

SUPPLEMENTARY REPORTS conditions which all produced good acquisition, no significant relationship was found between S-S interval and responses with IRTs that were briefer than the R-S interval. This finding is consistent with previous data (Anger, 1963) from well-trained animals, and is consistent with evidence that the major determinant of Sidman avoidance IRT distributions is the R-S rather than S-S interval. The nonsignificant decreasing level of responses with 0-2 sec. IRTs across S-S intervals of 1-5 sec. is probably due to the fact that repetitive responding after shock consists almost entirely of responses with brief IRTs. The nonsignificant increasing level of responses with IRTs of 2-20 sec. across S-S intervals of 1-5 sec. is the ascending limb of the significant inverted U-shaped function relating such responses to S-S intervals over the whole 1-20 sec. range. The consistent appearance of an ascending limb, as part of an inverted U-shaped function, at every intermediate 2-sec. IRT category from 2-20 sec. deserves attention, even though between group differences of Ss trained at S-S intervals of 1-5 sec. were not significant. It appears from these data that the condition with the longest S-S interval among several highly effective training conditions is the "best" Sidman avoidance training condition, from the standpoint of responses performed per shock delivered. REFERENCES
ANGER, D. The role of temporal discriminations in the reinforcement of Sidman avoidance behavior. J. exp. Anal. Behav., 1963, 6, 477-506. BLACK, A. H., & MOUSE, P. Avoidance learning in dogs without a warning stimulus. J. exp. Anal. Behav., 1961, 4,17-23. CAMPBELL, B. A., & TBGHTSOONIAN, R. Electrical and behavioral effects of different types of shock stimuli on the rat. J. comp. physiol. Psychol, 1958, 51, 185-192. HILL, H. E., FLANARY, H. G., KORNETSKT, C. H., & WIKLER, A. Relationship of electrically induced pain to the amperage and wattage of shock stimuli. /. din. Investig., 1962, 31, 464-472. SIDMAN, M. Avoidance conditioning with brief shock and no exteroceptive warning signal. Science, 1953, 118,157-158. SIDMAN, M. Reduction of shock frequency as reinforcement for avoidance behavior. J. exp. Anal. Behav., 1962, 5, 247-257. WEISSMAN, A. Nondiscriminated avoidance behavior in a large sample of rats. Psychol. Rep., 1962, 10, 591-600. (Received May 25, 1964)

sec. IRT responses were a nonsignificant increasing function of S-S interval (H = 1.5, dj = 2, p > .30), and IRTs greater than 20 sec. were a significant monotonic decreasing function of S-S interval (H = 10.8, df - 2, p < .01). Thus, the number of avoidance lever presses per shock delivered was an increasing function of S-S interval over the range from 1-5 sec. The 5-sec. Ss responded almost as often as the 1-sec. Ss, but received many fewer shocks. The effect of S-S interval on total shocks was nonmonotonic, but significant (H = 20.6, df = 4, p < .001). As Figure 1 indicates, the significance of this function was partly due to the restriction on the maximum total shocks in 6 hr. imposed on Ss with S-S intervals of 10 and 20 sec. Responding on the irrelevant lever was infrequent for all S-S interval groups (all means less than 400 responses). Most irrelevant responding occurred early in the session before avoidance response rates were stable. The observed behavior of all Ss in response to shock consisted of jumping and running movements. Every /S was observed for at least 20 successive shock presentations, and no instances of crouching or freezing during shock were ever observed. These observations often covered as many as 100 successive shocks, and were made at randomly selected time periods during experimental sessions. DISCUSSION The major findings of this experiment are (a) that reliable acquisition of Sidman avoidance lever pressing can be produced in 100% of an unselected group of rats under appropriate training conditions; (6) that first session response levels are a monotonically decreasing function of acquisition S-S interval; and (c) that the first session frequency of responses with IRTs of intermediate latency is an inverted U-shaped function of acquisition S-S interval. These data confirm Sidman's (1962) impression that acquisition during early exposure to avoidance contingencies is best at brief S-S intervals. The monotonic function observed is also consistent with Black and Morse's (1961) findings, with dogs in a shuttle box, that Sidman avoidance acquisition improved as S-S intervals approached escape contingencies (0sec. S-S interval). At the 1-, 3-, and 5-sec. intervals, under

Вам также может понравиться