Printer Friendly

Mechanism of stimulus classes formation in concurrent discriminations in rats.

There are many studies on a mechanism of stimulus classes formation in pigeons and rats. These studies make it clear that there are two main views on the mechanism of stimulus classes formation in pigeons and rats: One is a categorization process based on simple similarity between stimuli (Bhatt & Wasserman, 1989; Fersen & Lea, 1990; Vaughan & Herrnstein, 1987), the other is stimulus associations on the basis of reinforcement concordance (Delius, Ameling, Lea, & Staddon, 1995; Edwards, Jagielo, Zentall, & Hogan, 1982; Nakagawa, 1978, 1986, 1992, 1998; Urcuioli, Zentall, Jackson-Smith, & Steirn, 1989). Nakagawa (1978, 1986, 1992) has advocated a cue-associations theory on the mechanism of stimulus classes formation in concurrent discriminations in rats. According to Nakagawa (1978, 1986, 1992, 1998), during the original training, rats learn a connection between a positive stimulus and an approach response as well as a connection between a negative stimulus and an avoidance response for each discrimination task. They also form associations (i.e., cue associations) between the discriminative stimuli with the same response assignment during overtraining in concurrent discriminations. Nakagawa (1998) has made it clear that 120 trials of overtraining results in facilitating the formation of stimulus classes between the discriminative stimuli with the same response assignment, and that 240 trials of overtraining results in stimulus classes formation between the discriminative stimuli.

In spite of the reliability of stimulus classes formation between the discriminative stimuli in two concurrent discriminations in rats, a specific question remains. How do rats form cue associations or stimulus-stimulus associations (i.e., stimulus classes) between the discriminative stimuli with the same response assignment? That is, how do rats dissociate stimulus dimensions of two pairs of stimuli used in concurrent discriminations? This is a very important and ever fundamental issue in behavior analysis in studying the mechanism of stimulus classes formation in pigeons and rats. This problem has received far too little experimental attention in concurrent discrimination learning.

Zentall, Steirn, Sherburne, and Urcuioli (1991) have assumed a mechanism of the dissociation between stimulus dimensions of two pairs of stimuli as follows: Stimuli can have both unlearned and learned representations. An unlearned representation is one that is elicited by stimulus without learning and should be isomorphic with the stimulus [e.g., a red sample (R) could produce a "red" representation; a vertical sample (V), a "vertical" representation]. A learned representation is one that is established through experience. The learned representation might be arbitrary (e.g., R and V samples might both evoke an arbitrary representation "A"), or it could involve the unlearned representation of another stimulus. For example, in the many-to-one task, a V sample might produce a "red" representation, because both V and R samples were associated with the circle comparison (C) (p. 197). Although this assumption is speculative; it does, however, provide a possible explanation for stimulus classes formation.

One way to demonstrate that stimuli can have both unlearned and learned representations is to examine the effect of compounding discriminative stimuli after training on two concurrent discrimination tasks. According to this assumption advocated by Zentall et al. (1991), if the animals could form stimulus classes between the discriminative stimuli with the same response assignment in two concurrent discriminations (A+C-, B+D- for example), when they experienced a positive stimulus (A+) [or a negative stimulus (C-)] of the one discrimination task on a trial, they should then have not only the unlearned representation of the positive stimulus (A+) [(or the negative stimulus (C-)] but also the learned representation of another positive stimulus (B+) [or another negative stimulus (D-)] of the other discrimination task at the same time. Thus, when the animals were given both a new positive stimulus which had been comprised of two positive stimuli between the two discrimination tasks used in the original training (a positive compound stimulus) (A+B+) and a new negative stimulus comprised of two negative stimuli between the two discriminations used in the original training (a negative compound stimulus) (C-D-) after receiving enough overtraining, introduction of these new stimuli should then have little disruptive influence on the animals' subsequent discrimination performance. Because one stimulus has both the unlearned representation of it and the learned representation of another stimulus categorized in the stimulus class, the animals should be familiar with these two stimuli.

The present experiments were conducted to investigate directly whether or not each member of a stimulus class had both an unlearned representation of itself and a learned representation of another stimulus after forming stimulus classes between the discriminative stimuli in two concurrent discriminations. At the same time, the present research investigated how much overtraining was required for each member of the stimulus class to have both an unlearned and a learned representation. In order to achieve these aims, the present experiments conducted a limited parametric study of the training variable.

Experiment 1

The animals learned to discriminate concurrently two pairs of simple stimuli where the responses to one stimulus of each pair were rewarded and the responses to the other were not rewarded (A+C-, B+D-). One third of the animals were then concurrently overtrained in the two discrimination tasks for 10 days. Two thirds of the animals were then overtrained in just one of the two discrimination tasks for 10 days, either the A+C- task or the B+D- task for example. All the animals were tested four times with compound stimuli: after reaching the original learning's criterion (0), after being overtrained for 1 day (1), for 5 days (5), and for 10 days (10). The expectation according to both the assumption advocated by Zentall et al. (1991) and the findings of Nakagawa (1986, 1992, 1998) is that discrimination performance of the animals overtrained in two discrimination tasks should be superior to those of the animals overtrained in just one discrimination task on both the 5 and 10 test conditions.

Method

Subjects

Twenty-four experimentally naive male Sprague-Dawley rats were used. They were about 150 days old with an initial average body weight of 325 g. The animals were handled for 5 min a day for 12 days and were maintained on a daily 2-hr feeding schedule prior to the experiment. The amount of food in the daily ration was gradually reduced until the body weight of each animal was 80% of the baseline weight at the start of the experiment. Water was always available for the animals in their individual home cages. The animals were maintained on a 8-16-hr light:dark cycle, with light off at 10:00 p.m.

Apparatus

An automatic Y maze, shown in Figure 1, was used. The apparatus was painted medium grey inside and was illuminated throughout the experiment by a 10-W fluorescent lamp suspended 100 cm above the top of the apparatus. The start box measured 18 cm in height, 12 cm in width, and 15 cm in length. The distance from the start box to the bifurcation was 30 cm. The arms of the Y maze were 25 cm in length. Two end walls and side walls of the apparatus were medium-grey Plexiglas and the ceiling was clear Plexiglas. In the center of the end wall the start box had a food tray (5 cm x 4 cm x 1 m) into which a milk pellet was delivered from a feeder when the animals made a correct response. The end wall feeder when the animals made a correct response. The end wall each arm had one screen (12 cm square, 5 cm from the floor) and one response lever (4 cm square, 5 cm above the floor) at the center below each screen. A guillotine door opened and closed automatically to control access to the start box. That is, when the animals ran across a photo beam at the exit of the start box located 3 cm from the guillotine door, stimuli were rear-projected automatically onto screens; the animals ran down, pressed a correct response lever, returned to the start box, crossed a photoelectric gate located 7.5 cm from the end wall in the start box, and then the guillotine door closed automatically and after 10 sec the guillotine door opened automatically. The programming of events and data collection were carried out on line using a laboratory computer. Sound masking was provided by white noise from a blower fan (50 dB).

Stimuli

Training stimuli. Discriminative stimuli were rear-projected onto the screens by means of two Handy Cabin in-line projectors (Handy Cabin Ltd.). When the animals pressed a correct response lever, the discriminative stimuli disappeared. For a vertical-horizontal stripe discrimination, a vertically striped stimulus and a horizontally striped one were used, having alternating black and white lines, 1 cm in width. The triangle-circle discrimination used an equilateral triangle (with 10 cm sides) and a circle (with a diameter of 7.4 cm).

Test stimuli. Compound stimuli were used as test stimuli on test trials: A positive compound stimulus comprised both a positive stimulus of one discrimination task used in the original training and a positive stimulus of the other discrimination task. A negative compound stimulus was formed similarly. For example, if a triangle was a positive stimulus on the triangle-circle discrimination task and vertical stripes was a positive stimulus on the vertical-horizontal stripes discrimination task in the original training, a positive compound stimulus was comprised of a triangle on the right half of the slide and vertical stripes on the left half of the slide, and vice versa. A negative compound stimulus was comprised of a circle on the right half of the slide and horizontal stripes on the left half, and vice versa [ILLUSTRATION FOR FIGURE 2 OMITTED].

Procedure

Magazine training and shaping of lever press. The animals were concurrently given both magazine training and lever-press training in a Skinner box (15 cm high, 22.5 cm wide, and 15 cm long) with one screen measuring 5 cm long and 5 cm wide for 5 days until they pressed the lever at least 50 times for 30 min a day.

Pretraining. After completing both magazine training and lever-press shaping, the animals were given pretraining for 10 days prior to the beginning of the training phase until they pressed the lever at least 30 times per day on each side in the automatic Y maze. A medium-grey stimulus was rear-projected onto the screen during both shaping of lever press and pretraining.

Training. A trial in this experiment is defined as a response-stimulus sequence beginning when the animals start from the start box after opening the guillotine door, run down in the runway, press a correct response lever, and return to the start box.

Discrimination training (Original learning). The animals were trained concurrently to a criterion in the original learning for 24 trials a day on two discrimination tasks: triangle versus circle and vertical versus horizontal stripe. The criterion was 22 correct trials out of a possible 24 for each discrimination over 2 successive days combined. A self-correction training method was used in which, if the animals pressed the incorrect response lever, they were allowed to return to the choice point and to press the correct response lever. The positive and negative stimuli were counterbalanced. The order of trials with the two discrimination tasks followed four predetermined random sequences. The position of a positive stimulus also followed four predetermined random sequences. The animals were given one 45-mg milk pellet when they made a correct response. Intertrial interval was 10 sec.

After reaching the criterion, the animals were divided into three groups of eight animals: E, C-I, and C-II, matched with respect to the number of days to reach the criterion. The animals of Group E received further the same training on the two discrimination tasks for 10 days after reaching the original learning criterion. The animals of both Groups C-I and C-II received the training in just one of the two discrimination tasks for 10 days after reaching the original learning criterion. The animals of Group C-I received further training on just the triangle-circle task and those of Group C-II received further training on just the vertical-horizontal stripe task during overtraining.

Test. The animals of each group were tested four times with the compound stimuli task, after reaching the original learning criterion (0), after being overtrained for 1 day (1), for 5 days (5), and for 10 days (10). On each test, the animals were given five trials. They were given a reward on every trial regardless whether they pressed the correct or incorrect response lever. Other aspects of the conditions were the same as for the original training

Results

The group mean days-to-criterion on each discrimination task in Phase 1 training for each group are shown in Table 1. There was no indication of difference among these three groups in the rates at which they learned in Phase 1 training, and this observation was supported by statistical analysis. An ANOVA using group (E vs. C-I vs. C-II) and task (V-H vs. T-C) with repeated measure revealed that neither effects of group [F(2, 21) [less than] 1] and of task [F(1, 21) [less than] 1] nor interaction of group x task [F(2, 21) [less than] 1] were significant. The standard deviation of Table 1 appeared to be larger than the differences between the means. This was caused by the arrangement to equalize total number of days to criterion in the original learning for each group. The mean percentage of errors for all the animals during overtraining was 8%.
Table 1

Means and Standard Deviations of the Number of Days to Criterion in
Phase 1 Training in Experiment 1

 T-C H-V
Group / Task
 M SD M SD
E 66.38 9.51 65.50 9.75
C-I 67.88 7.49 64.25 6.20
C-II 66.50 13.68 63.75 15.60

Note. T = triangle stimulus, C = circle stimulus, H = horizontal
stripe stimulus, V = vertical stripe stimulus.


The results for each group in Phase 2 test are illustrated in Figure 3. An ANOVA using group (E vs. C-I vs. C-II) and degree of overtraining (0 vs. 1 vs. 5 vs. 10) with repeated measure was performed on the number of correct responses on each test phase, which revealed a statistically significant effect of group [F(2, 21) = 42.45, p [less than] .001] and a significant group x degree of overtraining interaction [F(6, 63): 2.97, p [less than] .05]. As the amount of overtraining increased, the number of correct responses of Group E increased significantly [F(3, 84) = 6.57, p [less than] .01], whereas those of both Groups C-I [F(3, 84) [less than] 1] and C-II [F(3, 84) = 1.64] neither increased nor decreased, which were at random level. There were significant differences in the number of correct responses on the 10 test condition [F(2, 84) = 15.89, p [less than] .001] and the 5 test condition [F(2, 84) = 5.00, p [less than] .01] among these three groups. A t test was used to compare the difference in the number of correct responses on the 10 and the 5 test conditions between these groups. The animals of Group E made significantly more correct responses on the 10 test condition than those of both Group C-I [t(14) = 4.34, p [less than] .01] and C-II [t(14) = 5.19, p [less than] .01] On the 5 test condition, the animals of Group E also made significantly more correct responses than those of both Group C-I [t(14) = 2.64, p [less than] .05 and Group C-II [t(14) = 3.24, p[less than] .01].

The animals of Group E did not make more errors on both the 10 [[x.sup.2](1) = 1.85] and the 5 [[x.sup.2](1) = 2.38] test conditions than during overtraining, whereas they made significantly more errors on both the 1 [[x.sup.2](1) = 19.42, p [less than] .01] and the 0 [[x.sup.2](1) = 39.69, p [less than] .01] test conditions than during overtraining. By contrast, performance of the animals of Group C-I on the 0 [[x.sup.2](1) = 33.95, p [less than] .01], the 1 [[x.sup.2](1) = 52.39, p [less than] .01], the 5 [[x.sup.2](1) = 28.67, p [less than] .01], and the 10 [[x.sup.2](1) = 45.82, p [less than] .01] test conditions, measured in terms of percentage of error, was significantly inferior to that during overtraining. Performance of the animals of Group C-II on the 0 [[x.sup.2](1) = 33.95, p [less than] .01], the 1 [[x.sup.2](1) = 19.42, p [less than] .01], the 5 [[x.sup.2](1) = 52.39, p [less than] .01], and the 10 [[x.sup.2](1) = 66.83, p [less than] .01] test conditions also, measured in terms of percentage of error, was inferior to that during overtraining

Discussion

With regard to the number of correct responses on the test, overtraining significantly improved discrimination performance in Group E, whereas it did not influence discrimination performance on the test in either Group C-I or Group C-II. In addition, for the animals of Group E, discrimination performance on both the 5 and 10 conditions were not inferior to that during overtraining, whereas those on both the 0 and 1 conditions were significantly inferior to that during overtraining. Furthermore, discrimination performance on both the 5 and 10 test conditions in either Group C-I or Group C-II were significantly inferior to that during overtraining. But there were not significant differences in the number of correct responses on both the 0 and 1 conditions among these three groups. These findings indicate that the introduction of compound stimuli has little disruptive influence on the animals' discrimination performance on the test after 10 days of overtraining in Group E, but it has a disruptive influence on the animals' discrimination performance on the test in either Group C-I or Group C-II. Nakagawa (1992, 1998) shows that the rats form stimulus classes between the discriminative stimuli with the same response assignment after overtraining. Taken together, the findings of this experiment make it clear that, after building up stimulus classes between the discriminative stimuli, each member of the stimulus class has both the unlearned representation of itself and the learned representation of another stimulus. Thus, these findings support the assumption advocated by Zentall et al. (1991).

The animals of Group E did not make more errors on the 5 test condition than during overtraining. This makes it clear that each member of the stimulus class begins to have both unlearned and learned representations after 120 overtraining trials. This finding is in line with the findings of Nakagawa (1998) with rats and Nakagawa (1985) with young children that rats begin to form stimulus classes between the discriminative stimuli with the same response after 120 overtraining trials in the concurrent discriminations and that young children begin to form stimulus classes between the discriminative stimuli after 12 overtraining trials in the concurrent discriminations.

Discrimination performance on the test for the animals in Group E was superior to those of the animals in either Group C-I or Group C-II on both the 5 and 10 test conditions, whereas there was no significant difference in discrimination performance among these three groups on both the 0 and 1 test conditions, which was at random level. The findings make it clear that overtraining in two concurrent discriminations results in the effective compounding of the discriminative stimuli, whereas overtraining in just one of the two discrimination tasks does not result in the effective compounding of the discriminative stimuli. They indicate that the effects of overtraining in two concurrent discriminations differ from those observed in a single discrimination. The findings are in line with the findings of Nakagawa (1992).

Alternatively, superiority of discrimination performance of the animals in Group E on the 5 and 10 test conditions to those of the animals in both Group C-I and Group C-II may be caused by strong response tendencies to all stimuli as a result of overtraining. Namely, in order to perform well with the relatively novel stimuli used on test sessions, the animals need to have strong appropriate response tendencies to all the constituent elements of the test stimuli. Only overtraining in the two discrimination tasks is likely to allow this. Poor discrimination performance on both the 5 and 10 conditions in both Group C-I and Group C-II was caused by receiving just one task during overtraining and by generalization decrement produced by the introduction of novel stimuli (i.e., compound stimuli).

Experiment 2

The results in Experiment 1 indicate that stimuli have both unlearned representation and learned representation after forming stimulus classes as a result of overtraining in two concurrent discriminations. An alternative possibility is, however, that the findings in Experiment 1 may be caused by strong response tendencies to all stimuli as a result of overtraining.

To test the above possibility further, Experiment 2 examined the effect of extinction on the compounding of the discriminative stimuli. If strong appropriate response tendencies to all stimuli as a result of overtraining would cause the effective compounding of the discriminative stimuli, when the animals were given extinction training after overtraining, appropriate response tendencies to all stimuli should then be perfectly extinguished with the consequence that the control of all the constituent elements of the test stimuli should be lost. Therefore, on the test trials after extinction training, the animals' discrimination performance should be at random level. Thus, the present experiment was conducted to investigate whether the effective compounding of the discriminative stimuli was caused by strong appropriate response tendencies to all the constituent elements of the test stimuli as a result of overtraining. In order to determine this, the animals were concurrently trained in two discrimination tasks similar to those described in Experiment 1. The animals were then overtrained on the two discrimination tasks. After completing overtraining, the animals were given extinction training in either both discrimination tasks or just one of the two discrimination tasks. After completing extinction training, the animals were given test trials with the compound stimuli. The expectation according to the alternative possibility (i.e., response tendencies account) is that discrimination performance on the test for the animals given extinction training on both discrimination tasks would be inferior to those of the animals given extinction training on just one discrimination task.

Method

Subjects

Fourteen experimentally naive Sprague-Dawley rats (12 females, 2 males) were used. They were about 130 days old with an initial average body weight of 155 g. All details of feeding schedule and apparatus were the same as in Experiment 1.

Stimuli

Training stimuli. For a white-black discrimination, a white stimulus and a black one were used. The same stimuli as in Experiment 1 were used for a vertical-horizontal stripe discrimination.

Test stimuli. Compound stimuli were used as a test stimulus on test trials: A positive compound stimulus comprised both a positive stimulus of one discrimination task used in the original training and a positive stimulus of the other discrimination and similarly for a negative compound stimulus.

Procedure

Magazine training and shaping of lever press. All details of magazine training, shaping of lever press, and pretraining were the same as in Experiment 1.

Discrimination training (Original learning). The animals were trained concurrently to a criterion in the original learning for 24 trials a day on two discrimination tasks: white versus black and vertical versus horizontal stripe. All details of discrimination training were the same as in Experiment 1.

After reaching the criterion of the original learning, the animals received the same training for an additional 10 days with the original training.

Extinction. After completing overtraining, the animals were divided into two groups: W and P, matched with respect to the number of days to criterion. The animals of both Groups W and P were given the same extinction training method as in Nakagawa (1986) to criterion. That is, for the animals in Group W, both doors were locked on all trials on both the white-black discrimination and the vertical-horizontal stripe discrimination tasks, although positive stimuli were still changed from side to side. Extinction continued for the animals in Group W until both positive and negative stimuli were chosen equally often over 12 successive trials on each task. For the animals in Group P, both doors were locked on all trials on the vertical-horizontal stripe discrimination, but the animals continued to receive the original discrimination training on all trials on the white-black discrimination task. Extinction continued for the animals in Group P until both positive and negative stimuli were chosen equally often over 12 successive trials on the vertical-horizontal discrimination.

Test. The animals of each group were tested seven times with the compound stimuli: After reaching the 50% level of correct responses on one training session in the original learning (50%), after reaching the 75% correct responses level (75%), after reaching the original learning criterion (0), after having received 24 overtraining trials (24), having 120 overtraining trials (120), having 240 overtraining trials (240), and after extinction (Ext.). The animals were given five trials on each test. They were given a reward on every trial, regardless of which lever they pressed.

Results

The group mean days-to-criterion on each discrimination task in Phase 1 training for each group are shown in Table 2. There was no evidence of a difference between these two groups in the rate at which they learned in Phase 1 training, and this observation was supported by statistical analysis. An ANOVA using group (W vs. P) and task (W-B vs. H-V) revealed that neither effects of group [F(1, 12) [less than] 1], and of task [F(1, 12) = 3.22] nor interaction of group x task [F(1,12) [less than] 1] were significant. The standard deviation of Table 2 appeared to be larger than the differences between the means. This was caused by the arrangement of equalizing total number of days to criterion in the original learning for each group. The mean percentage of errors of all the animals during overtraining was 7.7%.

The mean days-to-criterion in extinction were 2.43 (SD = 0.73) for Group W and 2.71 (SD = 1.48) for Group R The analysis by a t test revealed no significant effect of group [t(12) = 1.63].
Table 2
Means and Standard Deviations of the Number of Days to Criterion in
Phase 1 Training in Experiment 2

 B-W H-V
Group / Task
 M SD M SD

W 27.43 19.19 45.14 14.28
P 28.29 22.00 44.29 18.73

Note. B = black stimulus, W = white stimulus, H = horizontal stripe
stimulus, V = vertical stripe stimulus.


The results for each group in Phase 3 test are illustrated in Figure 4. An ANOVA using group (W vs. P) and degree of training (50% vs. 75% vs. 0 vs. 24 vs. 120 vs. 240) with repeated measure was performed on the number of correct responses on each test, which revealed a statistically significant effect of degree of training [F(5, 60) = 18.66, p [less than] .001]. A t test was used to compare the differences in the number of correct responses on each test condition. The animals of the two groups combined made more correct responses on the 240 test condition than the 120 test condition [t(13) = 2.73, p [less than] .05], the 24 test condition [t(13) = 5.55, p [less than] .01]. They made more correct responses on the 120 test condition than the 24 test condition [t(13) = 3.01, p [less than] .05], the 0 test condition [t(13) = 5.40, p [less than] .01], the 75% test condition [t(13) = 8.08, p [less than] .01], and the 50% test condition [t(13) = 6.76, p [less than] .01]. They made more correct responses on the 24 test condition than the 0 test condition [t(13) = 3.31, p [less than] .01], the 75% test condition [t(13) = 4.43, p [less than] .01], and the 50% test condition [t(13) = 3.31, p [less than] .01]. There were no significant differences in the number of correct responses among the three test conditions: the 0, the 75%, and the 50%.

The animals' performance on the 240 test condition, measured in terms of percentage of errors, was significantly superior to that during overtraining [[x.sup.2](1) = 8.68, p [less than] .01), whereas the animals' performance on the 120 test condition was not inferior to that during overtraining [[x.sup.2](1) = 1.43], and those on the 24 test condition was inferior to that during overtraining [[x.sup.2](1) = 6.77, p [less than] .01].

On the test after extinction, Group W made significantly more correct responses than Group P [t(12) = 7.70, p [less than] .01].

The animals' performance on the test after extinction in Group W, measured in terms of percentage of errors, was not inferior to that on the 240 test condition [[x.sup.2](1) = 1.50], whereas the animals' performance in Group P was significantly inferior to those on the 240 test condition [[x.sup.2](1) = 33.72, p [less than] .01], the 120 test condition [[x.sup.2](1) = 33.72, p [less than] .01], and the 24 test condition [[x.sup.2](1) = 4.97, p [less than] .05], which was at random level.

The mean percentage of correct responses on each trial on the test after extinction in both Group W and Group P are shown in Figure 5. An ANOVA using group (W vs. P) and trial (first vs. second vs. third vs. fourth vs. fifth) was performed on the number of correct responses on each trial, which revealed significant effect of group [F(1, 12) = 784.55, p [less than] .01]. That is, the animals in Group W made more correct responses on the test than the animals in Group P did. However, there was no significant difference in the number of correct responses on the first trial between Groups W and P. The animals in Group W made perfect discrimination performance on the second, the third, the fourth, and the fifth trials, whereas the animals in Group P performed at random level on each trial. That is, the animals in Group W made perfect discrimination performance after food reinforcement was reintroduced, whereas the animals in Group P did not improve after food reinforcement was introduced again.

Discussion

This experiment essentially replicated the pattern of results seen in Experiment 1 which indicated that overtraining in concurrent discriminations resulted in effective compounding of the discriminative stimuli. This indicates that each member of the stimulus class has both the unlearned representation and the learned representation after overtraining.

The main purpose of this study was to investigate the effective compounding of discriminative stimuli as a result of overtraining was caused by response tendencies to all the constituent elements of the test stimuli. In the present experiment, the animals' performances on the test after extinction in Group W were neither superior nor inferior to that on the 240 test condition, whereas the animals' performances on the test in Group P were significantly inferior to that on the 240 test condition, which were at random level. This result is not in line with the expectation according to the response tendencies account. Therefore, these findings make it clear that the effective compounding of the discriminative stimuli, as a result of overtraining seen in Experiment 1, is not caused by response tendencies to all the constituent elements of the test stimuli.

In addition, although there was no significant difference in discrimination performance on the first trial in the test after extinction training between Groups W and P, the animals in Group W made perfect discrimination performance on the second trial after food reinforcement was reintroduced, and so on the third, the fourth, the fifth trials. But the animals' performance in Group P was not improved after food reinforcement was reintroduced, which was at random level [ILLUSTRATION FOR FIGURE 5 OMITTED]. The additional findings are not caused by strong response tendencies to all stimuli as a result of overtraining, because there was significant difference in discrimination performance on test trials between Groups W and P. Indeed, if response tendencies to all stimuli are responsible for the difference in discrimination performance on test trials between these two groups, discrimination performance on test trials for Group W should then be inferior to that of Group P and be at random level. But the phenomenon was not observed. Thus, the additional findings indicate that both the unlearned and learned representations which each member of the stimulus class has acquired during overtraining remain intact after extinction, and they reinstate their functions when food reinforcement is introduced again on the first trial of the test. This finding is in line with the findings of Nakagawa (1986) that an effective interchange of the discriminative stimuli between two discrimination tasks is possible after extinction training. The extinction training used in the present experiment made the animals in Group P withhold both approach responses to a positive stimulus and avoidance responses to a negative stimulus of the one discrimination task (i.e., the extinguished task), whereas it made them continue to choose a positive stimulus and to avoid a negative stimulus of the other discrimination task (i.e., the unextinguished task) likely during overtraining. That is, the extinction training breaks up the relation of the unlearned and learned representations of one stimulus which have been acquired during overtraining for the animals of Group P. Thus, for the animals in Group P, each member of the stimulus class has only the unlearned representation of itself but not the learned representation of another stimulus after food reinforcement is introduced again. In contrast, the extinction training made the animals in Group W withhold both approach responses to each positive stimulus and avoidance responses to each negative stimulus of the two discrimination tasks. That is, the extinction training does not break up and rather maintains the relation of the unlearned and learned representations of one stimulus acquired during overtraining. Thus, for the animals of Group W, one stimulus has both the unlearned representation of itself and the learned representation of another stimulus formed during overtraining after food reinforcement is introduced again. The reinstatement of both the unlearned and learned representations relation by the reintroduction of food reinforcement is responsible for the superiority of the animals' discrimination performance on test trials in Group W to the animals' discrimination performance in Group P.

General Discussion

Two experiments have examined a premise of stimulus classes formation between the discriminative stimuli. That is, two experiments have examined whether stimuli have both the unlearned representation of itself and the learned representation of another stimulus via the same response or common shared response following the same consequence through experience using compounding of the discriminative stimuli procedure in two concurrent two-choice discriminations as a function of overtraining in rats. In Experiment 1, the animals were concurrently trained on two discrimination tasks to criterion. After reaching the criterion, they were overtrained in either two discrimination tasks (Group E) or just one of the two tasks (Group C-I or Group C-II) for 10 days. They were tested four times with compound stimuli: after reaching the criterion (0), after being overtrained for 1 day (1), for 5 days (5), and for 10 days (10). The animals of Group E made more correct responses on the 5 and 10 test conditions than either the animals of Group C-I or Group C-II. As the amount of overtraining increased, the number of correct responses of Group E increased, whereas those of both Groups C-I and C-II neither increased nor decreased, which were at random level. The effect of overtraining on discrimination performance on the test trials for the animals in Group E was replicated in Experiment 2.

These findings make it clear that overtraining facilitates compounding of the discriminative stimuli with the same response following the same consequence in two concurrent two-choice discriminations. Namely, they indicate that overtraining facilitates each member of the stimulus class having both the unlearned representation and the learned representation in two concurrent discriminations. These findings are basically in accordance with the findings of Nakagawa (1986, 1998). Thus, the results of the present two experiments produce several lines of evidence indicating that each member of the stimulus class has both the unlearned representation and the learned representation in two concurrent two-choice discriminations in rats. Furthermore, Experiment 2 shows that the animals' test performance after extinction in Group W is superior to the animals' test performance in Group P, and that the former is very excellent but the latter is poor and at random level. These findings provide stronger evidence that both the unlearned representation and the learned representation of stimuli acquired during overtraining remain intact after extinction and they reinstate their functions when food reinforcement is introduced again.

The main purpose of the present experiments was to investigate a premise of stimulus classes formation between the discriminative stimuli in two concurrent discriminations. The results of the two experiments reported here provide stronger evidences for the assumption that a mechanism of dissociation exists between discriminations dimensions advocated by Zentall et al. (1991). Thus, a premise of stimulus class formation between the discriminative stimuli in two concurrent discriminations is what one stimulus has both the unlearned representation of itself and the learned representation of the other stimulus with the same response assignment during enough overtraining in two concurrent discriminations in rats. That is, there are two premises for the animals to form stimulus classes in two concurrent discriminations: the one is that the animals receive enough overtraining (i.e., 240 overtraining trials) in two concurrent discriminations, the other is that one stimulus has both the unlearned representation of it and the learned representation of another stimulus. This conclusion is supported by the findings of Nakagawa (1992, 1998). Nakagawa (1992, 1998) showed that being concurrently overtrained in two discrimination tasks is necessary for the animals to form stimulus classes between the discriminative stimuli.

The results of the present experiments indicate that the number of correct responses on tests increases rapidly after 120 overtraining trials. Nakagawa (1998) found that the animals began to form stimulus classes between the discriminative stimuli after 120 overtraining trials. The findings of both the present experiments and Nakagawa (1998) make it clear that each member of the stimulus class begins to have both the unlearned representation and the learned representation in two concurrent discriminations after 120 overtraining trials, and at the same time, they begin to form stimulus classes between the discriminative stimuli with the same response following the same consequence. Thus, the process that one stimulus has both the unlearned and learned representations in two concurrent discrimination tasks is not necessarily in conflict with stimulus classes formation process advocated by the cue-associations theory (Nakagawa, 1986, 1992). Conversely, it seems that in nature both processes may always act synchronously.

The results of the experiments reported here make it clear that there are two premises for the animals to form stimulus classes between the discriminative stimuli in two concurrent discriminations in rats. Thus, the results of the present experiments may contribute to understanding a mechanism of stimulus associations in animals.

References

BHATT, R. S., & WASSERMAN, E. A. (1989). Secondary generalization and categorization in pigeons. The Journal of the Experimental Analysis of Behavior, 52, 213-224.

DELIUS, J. D., AMELING, M., LEA, S. E. G., & STADDON, J. E. R. (1995). Reinforcement concordance induces and maintains stimulus associations in pigeons. The Psychological Record, 45, 283-297.

EDWARDS, C. A., JAGIELO, J. A., ZENTALL, T. R., & HOGAN, D. E. (1982). Acquired equivalence and distinctiveness in matching to sample by pigeons. Journal of Experimental Psychology: Animal Behavior Processes, 8, 244-259.

FERSEN, L., VON, & LEA, S. E. G. (1990). Category discrimination with polymorphous feature. The Journal of the Experimental Analysis of Behavior, 54, 69-84.

NAKAGAWA, E. (1978). The effect of overtraining on discrimination learning in white rats (in Japanese). Japanese Journal of Psychology, 49, 70-77.

NAKAGAWA, E. (1985). Effects of overtraining on discrimination learning in young children - Examination of Shaeffer & Ellis's cue-cluster hypothesis (in Japanese). Japanese Journal of Educational Psychology, 33, 261-266.

NAKAGAWA, E. (1986). Overtraining, extinction and shift learning in a concurrent discrimination in rats. The Quarterly Journal of Experimental Psychology, 38B, 313-326.

NAKAGAWA, E. (1992). Effects of overtraining on reversal learning by rats in concurrent and single discriminations. The Quarterly Journal of Experimental Psychology, 44B, 37-56.

NAKAGAWA, E. (1998). Stimulus classes formation process in concurrent discriminations in rats as a function of overtraining. The Psychological Record, 48, 537-560.

URCUIOLI, P. J., ZENTALL, T. R., JACKSON-SMITH, P., & STEIRN, J. N. (1989). Evidence for common coding in many-to-one matching: Retention, intertrial interference, and transfer. Journal of Experimental Psychology: Animal Behavior Processes, 15, 264-273.

VAUGHAN, W., & HERRNSTEIN, R. J. (1987). Choosing among natural stimuli. The Journal of the Experimental Analysis of Behavior, 47, 5-16.

ZENTALL, T. R., STEIRN, J. N., SHERBURNE, L. M., & URCUIOLI, R J. (1991). Common coding in pigeons assessed through partial versus total reversals of many-to-one conditional and simple discriminations. Journal of Experimental Psychology: Animal Behavior Processes, 17, 194-201. imination tas
COPYRIGHT 1999 The Psychological Record
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Nakagawa, Esho
Publication:The Psychological Record
Date:Mar 22, 1999
Words:6863
Previous Article:Acquired equivalence of discriminative stimuli following two concurrent discrimination learning tasks as a function of overtraining in rats.
Next Article:Endogenous odor control of animal runway performance: generality, questions, and utility.
Topics:


Related Articles
The search for stimulus equivalence in nonverbal organisms.
Unreinforced conditional selection within three-choice conditional discriminations.
The effects of terminal-link stimulus arrangements on preference in concurrent chains.
Discrimination of "four" and "two" by pigeons.
Stimulus classes formation in concurrent discriminations in rats as a function of overtraining.
Stimulus equivalence and the blocking effect.
A factor affecting stimulus classes formation in concurrent discriminations in rats.
Acquired equivalence of discriminative stimuli following two concurrent discrimination learning tasks as a function of overtraining in rats.
Overtraining, extinction, and shift learning in matching-to-sample discriminations in rats.
Whole reversal versus partial reversal advantage effect on same-different discrimination in rats.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters