Psychological Reports, 1975, 37, 747-755. @ Psychological Reports 1975
R N TRANSITIONS: FRUSTRATION, INHIBITION OR AVERSION?l ROBERT E. PRYTULA, J. D W I G H T THOMAS, AND Middle Tennessee State University
JAMES H. PAYNE
AND STEPHEN F. DAVIS Austin Peay State University Sumwy.-Three groups of rats were trained and extinguished (Phase I ) in an alley, and retrained and extinguished (Phases 11 and 111) under RN, NR, and RR reward schedules. An exhaust system was operative in the alley in Phases I and 111 but inoperative in Phase 11. The results show that with the simple schedules employed, odors do not appear to mediate the running responses. Ss which received the RN schedule were generally less resistant to extinction than Ss receiving the other schedules. Following Phase 111, Ss in Group RN were retrained and extinguished under the N R sequence, and Ss in Group N R were shifted to the RN sequence. The results showed that if S received an NR sequence at some point in training later behavior was much more influenced by this schedule than the previous RN sequence. Also, it appeared that R N schedules are aversive, perhaps like the "Time-out" phenomenon.
Much theoretical and empirical controversy has centered around the n a m e and sequencing of rewarded ( R ) and nonrewarded ( N ) trials in relation to the partial reinforcement effect. Capaldi ( 1971) has maintained that the memory stimuli from the preceding nonrewarded trial ( s ) become conditioned to the running response on the subsequent rewarded trial ( s ) , thereby training the animal to run in the presence of the memory of nonreward. And with extinction, the previously established memory of nonreward and running in its presence are important in determining resistance to extinction. For Capaldi ( 1970, 197 1 ) the N R transition is the major determiner of the partial reinforcement effect. Sequences or schedules with zero NR transitions, i.e., an RN schedule, would not lead to an increase in resistance to exrinction since S is never reinforced in the presence of memory cues of nonreward. On the other hand, multiple-pellet RN transitions are the minimal conditions for establishing the effect, according to frustration theory (Amsel, Hug, & Surridge, 1968). Frustration theory asserts that with multiple-pellets on a rewarded trial an approach response and expectancy (r,) are established to cues in the goal-box. Subsequent disconfirmation of this expectancy by the following or nonrewarded trial leads to the development of frustraiion. Initially frustration cues ( s f ) lead to a disruption of the on-going response but with continued 'This research was supported by a Faculty Research Grant from Middle Tennessee State University to the first author. and a Tower Research Fund Grant from Austin Peav State university to the last author.. Reprint requests should be directed to the first authdr, Department of Psychology, Middle Tennessee State University, Murfreesboro, Tennessee 37130.
748
R. E. PRYTULA,
ET AL.
training sf cues become counterconditioned, now mediating an approach response (Amsel, 1972 ) . The purpose of the present study was threefold: ( 1 ) to assess the effects of presenting animals RN, NR, and RR transitions when not embedded in more complex schedules; ( 2 ) to assess the effects o f the above transitions under odorminimized and standard laboratory conditions. Several studies (Seago, Ludvigson, & Remley, 1970; Prytula, Bridges, Anderson, & Hayes, 1972; Prytula, Bridges, R: Cox, 1973) have shown that rat-generated olfactory cues may be more important behavior determinants than memory, especially when the reinforcement schedule is complex. However, it is distinctly possible that odor cues may not be as vital and/or utilized as fully under simpler schedules such as those employed in the present study. And, ( 3 ) to assess the effects of the initial schedule upon later retraining and extinction under the opposite schedule.
METHOD Subjects Ss were 26 male rats purchased from [he Simonsen Laboratories, Gilroy, California. At the beginning of training Ss were approximately 9 0 days old. Ss were individually caged and fed 1 3 gm of Purina lab chow on a 22Y'-hr. deprivation schedule 2 wk. prior to and throughout the experiment. Water was continuously available in the home cage. Apparatus The apparatus was a gray wooden runway 152.4 crn long, 9.16 cm wide, and 16.51 cm high. A 30.48-cm start section was separated from an 88.90-cm run section by a frosted Plexiglas door. In the goal section, a second guillotine door separated the run section from a 30.48-cm goal box. Lifting the start door activated a Standard Electric timer, and interrupting a photobearn located 20.39 cm beyond [he start door stopped the first timer (start time) and activated a second timer. Interrupting a second photobearn located 60.96 un beyond the first photobeam stopped the second timer ( r u n time) and started a third timer. Interrupting a third photobenm located 30.48 cm beyond the second photobeam and 2.54 cm in front of a recessed goal cup stopped the third timer (goal time). The entire apparatus was floored with %-in. hardware cloth and was covered with clear Plexiglas lids. Running underneath the entire runway was a U-shaped, metal tunnel forming an odor-exhaust duct. Located and recessed (22.86 c m ) beyond the goal segment was a small, exceptionally quiet fan continuous with the tunnel forming the odor-exhaust system. Two days prior to the beginning of the experimental phases, each S received two pretraining trials. Pretraining consisted of 5 min. of free exploration with 20 ,045-gm. Noyes pellets available in both goal area and home cage. During these trials all photoe!ectric and exhaust equipment was operative. Prior to pretraining Ss were randomly assigned to one of three groups; RN ( n = g ) , NR ( n = 9 ) , and RR ( n = 8 ) . Group R N received one rewarded trial followed 1 min. later by a noncewarded trial. Group NR's first trial was a nonrewarded trial followed 1 min. later by a rewarded trial; and Group R R received rwo rewarded trials separated by the same l-min. intertrial interval. During all phases of [he experiment each S
RN TRANSITIONS
749
followed itself with no other S intervening between the trial administrations. Food reward consisted of 20 ,045-gm. Noyes pellets with S being removed immediately after consuming the last pellet. Nonreward confinement was 30 sec. during both acquisition and extinction. The experiment consisted of four phases. During Phase I acquisition each S received two trials a day for 12 days according to the reward-nonreward conditions described above. Five days of extinction followed acquisition training, Ss were given two trials per day. During rhis phase the exhaust system was operative for each S in all groups while S was in the runway and for a 2-min. interval between Ss. Following a seven-day period, Phase I1 was initiated. During this period Ss were kept on the maintenance diet and handled each day. Phase I1 consisted of a retraining phase, seven days at two trials per day and an extinction phase of five days. All other conditions were identical to Phase I with the exception that the exhaust fan was inoperative. Following a seven-day interval, Ss in each group were retrained and extinguished for a third time (Phase 111). Conditions were identical to those of Phase I[ with the exception that the exhaust fan was again operarive as in Phase I. At the conclusion of Phase 111, Ss were kept on the maintenance schedule for a 2-wk. period. Following rhis interval Phase IV was initiated.' Conditions were identical to those of Phase I with the following exceptions: (1) Group RN's schedule was shifted to an NR sequence and Group N R was sh~fredto an RN sequence. Group R R remained unchanged, and ( 2 ) acquisition consisted of a 10-day period. During this phase, as in Phase 1, the exhaust system was operative. During all phases the running order of S within a group was varied each day, and the running order of groups across days was counterbalanced. After all Ss within any group had been run, a 15-min. interval elapsed before the next group was run. During this 15-min. period, with [he exception of Phase 11, the exhausr fan was operative.
RESULTS Start, run, and goal times for terminal acquisition (last three days) and 1 ) scores prior to analysis. These extinction were transformed into log,, (Xi transformed scores were subjected to analysis of variance with the reward schedule (groups) as the between factor and trials within days as the repeated measures factor. Subsequent comparisons of significant main effects and interactions were performed by Tukey's procedure.
+
Acquisition: Phases I, 11, a;& 111 The results of Phases I and I1 were virtually identical to those for Phase 111, therefore, only the latter are presented. The only difference was that under the nonexhaust conditions of Phase 11, the terminal level of Groups RR and RN was higher (slower approach rate) than under the exhaust conditions, i.e., Phases I and 111. With retraining under the exhaust condition, Ss in Group RN ran slower than Ss in Group RR ( p < .01) in the start segment, and slower than Ss in Group NR (ps < .01) in the run and goal segments (see Fig. 1 ). Analyses of the Groups X Trials interaction showed that Ss in Group NR started and ran 'Due to the boredom i n running a lengthy study, a second experienced E ran Phase 1V. Statistical analyses indicated that this was a nonsignificant factor.
R. E. PRYTLJLA, ET AL.
l.2. l.1'
-
1.0.
W RH
1
START
ACOUISITION Ill
0-43
NR
1.3.
RUN
1.2-
*
EXlINCTlON III
RR
1.1
Y
u 1.0-
yr
-5 2*
.9-
.a-
FIG. 1. Group mean latencies during Phase 111: Acquisition, Days 30 to 36; Phase 111: Extinction, Days 37 to 41
.2-
1
-
1.3-
GOAL
1.21.01.1 .9
-
-
.21
1
1
1
1
1
1
30 31 32 33 34 35
3k 37 ?a & 4 b I: 1
significantly slower on Trial 1 than Groups RR (9s < .05) or RN (ps < .OS; see Fig. 2 ) . On Trial 2, Ss in Group R N ran significantly slower in all sections than Groups NR (ps < .01) or RR (ps < .01). Comparisons of the Trials effect within each group showed that the slow-down observed in Group RN from Trial 1 to Trial 2, and the increase in Group NR was significant in all sections of the runway (all ps < .01; see Fig. 2 ) . Statistical support for these statements can be summarized briefly: Phase 111, scart (Groups, P2.z3 = 5.95, p < .05; Groups X Trials, P2,Z3 = 18.27, 9 < ,001; Days, P 2 . 4 8 = 4.85, P < .05; and Groups X Days X Trials, F4.40 = 4.09, p < ,011 ; run (Groups F2,03 = 8.53,
RN TRANSITIONS ACOUlSlTlON PHASE Ill START
PUN
GOAL
FIG. 2. Groups x Trials mean cies for Acquisition: Phase I11
TRIALS
P < F2,?3
.005; Groups X Trials, F2.23 = 33.46, p < .001); and goal (Groups < .001; and Groups X Trials, F2,23 = 29.51, p < .001).
= 10.37, p
Extinction: Phases I, 11, and I I I
Under the exhaust condition of Phase 111, Ss in Groups RN and RR ap.01; proached the goal at a significantly slower race than Ss in Group NR (ps see Fig. 1 ) . Analysis of the Groups X Trials interaction showed that during Phase 111, Ss in Group NR on Trial 1 ran significantly slower in the start and goal sections than Group RN (9s < .05), and significantly slower than both Groups RR and RN (ps < .01) in the run section. On Trial 2, Ss in Groups RR and R N ran signifiacntly slower in all three sections of the alley than Ss in Group NR (ps < .01); ailso Ss in Group R N approached the goal at a significantly slower rate than Group RR ( p < .01; see Fig. 3). Comparisons of the Trials effect within each group showed that Ss in Groups RN and RR significantly slowed down in all sections on Trial 2 (ps < .01); while for Ss in Group NR the speed increase from Trial 1 to Trial 2 (ps < .01) was significant in all
R N TRANSITIONS: FRUSTRATION, INHIBITION OR AVERSION?l ROBERT E. PRYTULA, J. D W I G H T THOMAS, AND Middle Tennessee State University
JAMES H. PAYNE
AND STEPHEN F. DAVIS Austin Peay State University Sumwy.-Three groups of rats were trained and extinguished (Phase I ) in an alley, and retrained and extinguished (Phases 11 and 111) under RN, NR, and RR reward schedules. An exhaust system was operative in the alley in Phases I and 111 but inoperative in Phase 11. The results show that with the simple schedules employed, odors do not appear to mediate the running responses. Ss which received the RN schedule were generally less resistant to extinction than Ss receiving the other schedules. Following Phase 111, Ss in Group RN were retrained and extinguished under the N R sequence, and Ss in Group N R were shifted to the RN sequence. The results showed that if S received an NR sequence at some point in training later behavior was much more influenced by this schedule than the previous RN sequence. Also, it appeared that R N schedules are aversive, perhaps like the "Time-out" phenomenon.
Much theoretical and empirical controversy has centered around the n a m e and sequencing of rewarded ( R ) and nonrewarded ( N ) trials in relation to the partial reinforcement effect. Capaldi ( 1971) has maintained that the memory stimuli from the preceding nonrewarded trial ( s ) become conditioned to the running response on the subsequent rewarded trial ( s ) , thereby training the animal to run in the presence of the memory of nonreward. And with extinction, the previously established memory of nonreward and running in its presence are important in determining resistance to extinction. For Capaldi ( 1970, 197 1 ) the N R transition is the major determiner of the partial reinforcement effect. Sequences or schedules with zero NR transitions, i.e., an RN schedule, would not lead to an increase in resistance to exrinction since S is never reinforced in the presence of memory cues of nonreward. On the other hand, multiple-pellet RN transitions are the minimal conditions for establishing the effect, according to frustration theory (Amsel, Hug, & Surridge, 1968). Frustration theory asserts that with multiple-pellets on a rewarded trial an approach response and expectancy (r,) are established to cues in the goal-box. Subsequent disconfirmation of this expectancy by the following or nonrewarded trial leads to the development of frustraiion. Initially frustration cues ( s f ) lead to a disruption of the on-going response but with continued 'This research was supported by a Faculty Research Grant from Middle Tennessee State University to the first author. and a Tower Research Fund Grant from Austin Peav State university to the last author.. Reprint requests should be directed to the first authdr, Department of Psychology, Middle Tennessee State University, Murfreesboro, Tennessee 37130.
748
R. E. PRYTULA,
ET AL.
training sf cues become counterconditioned, now mediating an approach response (Amsel, 1972 ) . The purpose of the present study was threefold: ( 1 ) to assess the effects of presenting animals RN, NR, and RR transitions when not embedded in more complex schedules; ( 2 ) to assess the effects o f the above transitions under odorminimized and standard laboratory conditions. Several studies (Seago, Ludvigson, & Remley, 1970; Prytula, Bridges, Anderson, & Hayes, 1972; Prytula, Bridges, R: Cox, 1973) have shown that rat-generated olfactory cues may be more important behavior determinants than memory, especially when the reinforcement schedule is complex. However, it is distinctly possible that odor cues may not be as vital and/or utilized as fully under simpler schedules such as those employed in the present study. And, ( 3 ) to assess the effects of the initial schedule upon later retraining and extinction under the opposite schedule.
METHOD Subjects Ss were 26 male rats purchased from [he Simonsen Laboratories, Gilroy, California. At the beginning of training Ss were approximately 9 0 days old. Ss were individually caged and fed 1 3 gm of Purina lab chow on a 22Y'-hr. deprivation schedule 2 wk. prior to and throughout the experiment. Water was continuously available in the home cage. Apparatus The apparatus was a gray wooden runway 152.4 crn long, 9.16 cm wide, and 16.51 cm high. A 30.48-cm start section was separated from an 88.90-cm run section by a frosted Plexiglas door. In the goal section, a second guillotine door separated the run section from a 30.48-cm goal box. Lifting the start door activated a Standard Electric timer, and interrupting a photobearn located 20.39 cm beyond [he start door stopped the first timer (start time) and activated a second timer. Interrupting a second photobearn located 60.96 un beyond the first photobeam stopped the second timer ( r u n time) and started a third timer. Interrupting a third photobenm located 30.48 cm beyond the second photobeam and 2.54 cm in front of a recessed goal cup stopped the third timer (goal time). The entire apparatus was floored with %-in. hardware cloth and was covered with clear Plexiglas lids. Running underneath the entire runway was a U-shaped, metal tunnel forming an odor-exhaust duct. Located and recessed (22.86 c m ) beyond the goal segment was a small, exceptionally quiet fan continuous with the tunnel forming the odor-exhaust system. Two days prior to the beginning of the experimental phases, each S received two pretraining trials. Pretraining consisted of 5 min. of free exploration with 20 ,045-gm. Noyes pellets available in both goal area and home cage. During these trials all photoe!ectric and exhaust equipment was operative. Prior to pretraining Ss were randomly assigned to one of three groups; RN ( n = g ) , NR ( n = 9 ) , and RR ( n = 8 ) . Group R N received one rewarded trial followed 1 min. later by a noncewarded trial. Group NR's first trial was a nonrewarded trial followed 1 min. later by a rewarded trial; and Group R R received rwo rewarded trials separated by the same l-min. intertrial interval. During all phases of [he experiment each S
RN TRANSITIONS
749
followed itself with no other S intervening between the trial administrations. Food reward consisted of 20 ,045-gm. Noyes pellets with S being removed immediately after consuming the last pellet. Nonreward confinement was 30 sec. during both acquisition and extinction. The experiment consisted of four phases. During Phase I acquisition each S received two trials a day for 12 days according to the reward-nonreward conditions described above. Five days of extinction followed acquisition training, Ss were given two trials per day. During rhis phase the exhaust system was operative for each S in all groups while S was in the runway and for a 2-min. interval between Ss. Following a seven-day period, Phase I1 was initiated. During this period Ss were kept on the maintenance diet and handled each day. Phase I1 consisted of a retraining phase, seven days at two trials per day and an extinction phase of five days. All other conditions were identical to Phase I with the exception that the exhaust fan was inoperative. Following a seven-day interval, Ss in each group were retrained and extinguished for a third time (Phase 111). Conditions were identical to those of Phase I[ with the exception that the exhaust fan was again operarive as in Phase I. At the conclusion of Phase 111, Ss were kept on the maintenance schedule for a 2-wk. period. Following rhis interval Phase IV was initiated.' Conditions were identical to those of Phase I with the following exceptions: (1) Group RN's schedule was shifted to an NR sequence and Group N R was sh~fredto an RN sequence. Group R R remained unchanged, and ( 2 ) acquisition consisted of a 10-day period. During this phase, as in Phase 1, the exhaust system was operative. During all phases the running order of S within a group was varied each day, and the running order of groups across days was counterbalanced. After all Ss within any group had been run, a 15-min. interval elapsed before the next group was run. During this 15-min. period, with [he exception of Phase 11, the exhausr fan was operative.
RESULTS Start, run, and goal times for terminal acquisition (last three days) and 1 ) scores prior to analysis. These extinction were transformed into log,, (Xi transformed scores were subjected to analysis of variance with the reward schedule (groups) as the between factor and trials within days as the repeated measures factor. Subsequent comparisons of significant main effects and interactions were performed by Tukey's procedure.
+
Acquisition: Phases I, 11, a;& 111 The results of Phases I and I1 were virtually identical to those for Phase 111, therefore, only the latter are presented. The only difference was that under the nonexhaust conditions of Phase 11, the terminal level of Groups RR and RN was higher (slower approach rate) than under the exhaust conditions, i.e., Phases I and 111. With retraining under the exhaust condition, Ss in Group RN ran slower than Ss in Group RR ( p < .01) in the start segment, and slower than Ss in Group NR (ps < .01) in the run and goal segments (see Fig. 1 ). Analyses of the Groups X Trials interaction showed that Ss in Group NR started and ran 'Due to the boredom i n running a lengthy study, a second experienced E ran Phase 1V. Statistical analyses indicated that this was a nonsignificant factor.
R. E. PRYTLJLA, ET AL.
l.2. l.1'
-
1.0.
W RH
1
START
ACOUISITION Ill
0-43
NR
1.3.
RUN
1.2-
*
EXlINCTlON III
RR
1.1
Y
u 1.0-
yr
-5 2*
.9-
.a-
FIG. 1. Group mean latencies during Phase 111: Acquisition, Days 30 to 36; Phase 111: Extinction, Days 37 to 41
.2-
1
-
1.3-
GOAL
1.21.01.1 .9
-
-
.21
1
1
1
1
1
1
30 31 32 33 34 35
3k 37 ?a & 4 b I: 1
significantly slower on Trial 1 than Groups RR (9s < .05) or RN (ps < .OS; see Fig. 2 ) . On Trial 2, Ss in Group R N ran significantly slower in all sections than Groups NR (ps < .01) or RR (ps < .01). Comparisons of the Trials effect within each group showed that the slow-down observed in Group RN from Trial 1 to Trial 2, and the increase in Group NR was significant in all sections of the runway (all ps < .01; see Fig. 2 ) . Statistical support for these statements can be summarized briefly: Phase 111, scart (Groups, P2.z3 = 5.95, p < .05; Groups X Trials, P2,Z3 = 18.27, 9 < ,001; Days, P 2 . 4 8 = 4.85, P < .05; and Groups X Days X Trials, F4.40 = 4.09, p < ,011 ; run (Groups F2,03 = 8.53,
RN TRANSITIONS ACOUlSlTlON PHASE Ill START
PUN
GOAL
FIG. 2. Groups x Trials mean cies for Acquisition: Phase I11
TRIALS
P < F2,?3
.005; Groups X Trials, F2.23 = 33.46, p < .001); and goal (Groups < .001; and Groups X Trials, F2,23 = 29.51, p < .001).
= 10.37, p
Extinction: Phases I, 11, and I I I
Under the exhaust condition of Phase 111, Ss in Groups RN and RR ap.01; proached the goal at a significantly slower race than Ss in Group NR (ps see Fig. 1 ) . Analysis of the Groups X Trials interaction showed that during Phase 111, Ss in Group NR on Trial 1 ran significantly slower in the start and goal sections than Group RN (9s < .05), and significantly slower than both Groups RR and RN (ps < .01) in the run section. On Trial 2, Ss in Groups RR and R N ran signifiacntly slower in all three sections of the alley than Ss in Group NR (ps < .01); ailso Ss in Group R N approached the goal at a significantly slower rate than Group RR ( p < .01; see Fig. 3). Comparisons of the Trials effect within each group showed that Ss in Groups RN and RR significantly slowed down in all sections on Trial 2 (ps < .01); while for Ss in Group NR the speed increase from Trial 1 to Trial 2 (ps < .01) was significant in all
