Printer Friendly

Superstitious behavior and response stereotypy prevent the emergence of efficient rule-governed behavior in animals.

Concept learning is an ability that is acknowledged in humans. Whether other animals can learn concepts has been a debate for many decades (e.g., Hulse, Fowler, & Honig, 1978; Kendrick, Rifling, & Denny, 1986; Roitblat, Bever, & Terrace, 1984; Spear & Miller, 1981).

According to Slavin (1997), "A concept is an abstract idea that is generalized from specific examples" (p. 251). Several researchers have provided experimental criteria for evidence of concept learning in humans and animals. Keller and Schoenfeld (1950) proposed that "Generalization within classes and discrimination between classes - this is the essence of concepts" (p. 155). Wasserman, Kiedinger, and Bhatt (1988) have expanded this definition to include cases wherein "an organism exhibits conceptual behavior if it responds similarly to old and new members of one class of stimuli and if it responds differently to old and new members of other classes of stimuli" (p. 235).

Evidence of concept learning does not necessarily require errorless transfer of performance to novel problems. A "savings in learning" new problems has been suggested to be sufficient evidence of concept learning (Premack, 1978). Others have suggested that concept learning is demonstrated when the organism's behavior appears to be under the control of a rule involving "same/different" judgments (Fujita, 1983; Herrnstein, Loveland, & Cable, 1976; Pisacreta, 1990).

There are two variables that can account for the failure of animals (and humans) to exhibit concept learning: superstition and response stereotypy. Experimentally, superstition has been defined as behavior controlled by a belief that certain behaviors contribute to reinforcement when, in fact, they are not necessary (Skinner, 1948). Behavior that is under the control of a superstitious rule typically develops into stereotyped response chains (Heltzer & Vyse, 1994). That is, given multiple ways of solving a problem, the subject emits the same response sequence over and over.

Three types of procedures have been used to study superstition and stereotypy (S&S): Schedule-Induced S&S, Spatial-Problems-Induced S&S, and Visual-Problems-Induced S&S. The first involves the use of reinforcement schedules that produce schedule-induced stereotypy. Skinner (1948) presented pigeons with food at regular intervals. No particular behavior was required nor did any behavior influence the food presentations. The six pigeons developed stereotyped behavior (e.g., turning in circles, head bobbing) that they emitted between food deliveries. Leander, Lippman, and Meyer (1968) reported that humans working on a fixed interval schedule would emit high rates of behavior (rather than a scallop) and the subjects reported that reinforcement depended on the number of responses emitted rather than time. Catania and Cutts (1963) trained college students on a concurrent variable interval (VI)-extinction schedule. Although only responses on the VI button produced reinforcers, the subjects reliably made responses on the extinction button. When interviewed, the subjects claimed that responses on the extinction button were required in order for the other (VI) button to pay off.

Spatial-Problems-Induced S&S typically require the subject to "move" a stimulus within a matrix. Extending on the work of Vogel and Annau (1973), Schwartz (1980) presented pigeons with two pecking keys and a 5 x 5 matrix of lights. Each trial began with the top left lamp illuminated. A peck on the left response key extinguished the lamp and lit the one to its right. A response on the right response key extinguished the lamp and lit the one below it. Reinforcement was provided after the bird had "moved" the light from the upper left corner of the matrix to the lower right corner. This required four responses to each pecking key in any order, for example, LLLLRRRR, RLRLRLRL. A fifth response to a pecking key terminated the trial without reinforcement (a penalty). Of the 70 possible sequences, the birds typically emitted one sequence during the majority of the trials.

It can be argued that the penalty contingency encouraged the formation of a stereotyped response sequence. However, in a related experiment, Pisacreta (1982a) presented pigeons with six to nine lighted keys (across conditions). A peck on a key darkened it. Reinforcement was presented when all the keys were extinguished. There was no penalty in this procedure. Still, of the 720 to 362,880 response sequences possible, the birds emitted less than 20 sequences (.0001 to 6% of the possible sequences), and, across conditions, one sequence occurred during the majority of the trials.

Schwartz (1982) used his procedure described above with human subjects. They produced more stereotypy than the pigeons did. Further, the subjects claimed that the sequence that they used was necessary in order to receive reinforcers (i.e., a superstitious belief). More recently, Heltzer and Vyse (1994) used a computer monitor to present college students with a 5 x 5 matrix of boxes. A circle was present in the upper left corner box. The subjects moved the circle from the upper left box to the lower right box by pushing two keys on the computer keyboard. Again, the subjects produced stereotypy and claimed that a certain sequence was necessary in order to succeed. Using a 6 x 6 matrix, Lopatto and Brown (1994) reported similar results. One procedural commonality that these experiments have is that they typically allow many response sequences to produce reinforcers, and one sequence is no more or less efficient than any other. Therefore, there is no explicit benefit in producing response variability across trials.

Visual-Problems-Induced S&S experiments require the subject to make responses that change the nature of the stimuli presented to them, rather than the location of the stimuli. Also, although there are several strategies that will yield reinforcement, some are much more efficient than others. Hence, if stereotypy develops, it can cost the subject hundreds, or thousands, of additional responses to obtain the same number of reinforcers.

For example, Pisacreta (1982b) trained pigeons to peck five illuminated keys. At trial onset, the keys presented one of two colors, for example, four red keys and a blue key. A peck to a red key changed the color to blue. Another peck replaced blue with green. A third peck changed the key back to red. Grain was presented when all of the keys showed the same color. The most efficient way to respond was to peck only the key(s) that did not match the majority of the keys. In the present example, two pecks to the blue key would produce five red keys. The birds could make all keys blue (four pecks) or green (nine pecks). Hence, repeatedly making the keys exhibit a particular stimulus, trial after trial, is very inefficient. Operating under a rule, "make them all match the dominant color" would result in response variability and the least number of responses per reinforcers. The pigeons matched a particular stimulus during 60%-100% of the trials and produced efficiencies of less than 50%. Increasing the number of stimuli available on each key to 12 did not reduce stereotypy. One explanation for their failure to use efficient strategies would be that pigeons lack the cognitive ability to learn rules. However, there is an abundant literature that demonstrates that pigeons can operate under a rule and make "same-different" judgments (e.g., Fujita, 1983; Herrnstein et al., 1976; Pisacreta, 1996, Wasserman et al., 1988). Another possibility is that procedures that reinforce stereotypy prevent the development of rule-governed behavior.

In a related paper, Pisacreta, Redwood, and Witt (1987) trained pigeons with a 3 x 3 matrix of pecking keys. At trial onset, the lower three keys presented red (S+) while the upper six keys were illuminated with green (S-). Every t sec, the red hues "moved" up one horizontal row. A peck to a red key darkened it as well as any key above or below it, that is, the left, center, or right column of keys. In order to produce a reinforcer, the birds had to peck each of the three red keys before any of them had been present on the upper level key for t sec. If the bird failed, a 3-s time-out was presented followed by a repeat of the same trial. Each of the birds produced a stereotyped sequence that they emitted during the majority of trials. For example, one bird pecked the three red keys from left to right. Changing the stimulus array at trial onset did not affect the sequence emitted. If a trial presented two red keys at the bottom of the matrix but the third was in the upper right corner, the bird did not peck that key first. It continued to peck the keys from left to right. This behavior resulted in frequent time-outs. The birds frequently produced less than 50% efficiencies. That is, they had to complete more than 60 trials in order to obtain 30 reinforcers. Again, one could argue that the rule was beyond the birds' cognitive abilities or that stereotypy prevented a rule from controlling behavior.

In a recent paper, Lee (1996) presented students with five computer icons. The student's task was to press the mouse button in order to change the icons. The students earned points if they changed the icons until all five matched and then pushed the up arrow key. Some trials were programmed to not allow several button depressions to change an icon. Although it was not required, several subjects frequently moved the mouse cursor to certain icons before depressing the button. Moving the cursor had no programmed consequence. However, the subjects believed that placing the cursor in certain positions was necessary in order to make the icons change.

Collectively, these studies suggest that the tendency to form superstitious associations between environmental events and behavior is a robust phenomenon in both animals and humans, as evidenced by the ease with which stereotyped behavior emerges, even if it is detrimental to success. Further studies that compare the performance of animals and humans working under comparable conditions should help illuminate our understanding of these phenomena.

EXPERIMENT 1

The purpose of Experiment 1 was to train humans on a computerized version of the procedure that Pisacreta (1982b) had used with pigeons. If humans, with their concept learning abilities, failed to use economical solutions (i.e., make all the stimuli the same with the least number of responses) because of the tendency to develop superstitious rules and stereotypy, then perhaps the failure of the birds on a similar task was caused more by the procedure's tendency to produce superstition and stereotypy rather than a cognitive deficit. Several variations of the Pisacreta (1982b) procedure were explored in order to study which conditions were likely to produce less, or more, stereotypy.

Method

Subjects

One hundred and thirty-one college students agreed to serve as subjects in the experiment. They were individually recruited as they passed by the research lab. In order to avoid using the same subjects twice, subjects were asked if they had participated in the experiment before. Each of the subjects participated in only one of the conditions described below.

Apparatus

An ATARI 130XE computer, 1050 disc drive, 1025 printer, and an AMDECK 500, 13-in color monitor were located on a table in a 3-m x 4-m room. The keyboard was placed directly in front of the monitor.

Procedure

Six different conditions were run during Experiment 1. They are summarized in Table 1.

Condition 1 - Three Letters: A, B, C.

Twenty-four subjects participated. Each naive subject was seated in front of the keyboard and was informed that the computer would explain what to do. The experimenter typed in the word RUN to initiate the session and then left the room.

The monitor displayed the following message:

WELCOME AND THANK YOU FOR COMING IN TODAY. IN A MOMENT, YOU WILL WORK ON AN EASY TASK. SO PLEASE RELAX AND DON'T BE NERVOUS. WHEN YOU ARE READY, PUSH ANY KEY TO CONTINUE.

[TABULAR DATA FOR TABLE 1 OMITTED]

The letters were .79 cm tall. They were white letters projected on a blue background. The following messages were presented after the subject pressed a key:

THIS IS NOT AN IQ TEST. YOU ARE NOT BEING TIMED ND THERE ARE NO TRICKS. SO PLEASE TRY TO ENJOY IT. PUSH ANY KEY TO CONTINUE.

After pressing a key, the display read:

DURING EACH TRIAL, YOU WILL BE PRESENTED WITH FIVE ALPHABET LETTERS. THE NUMBERS 1-5 WILL APPEAR UNDER THEM. PUSH ANY KEY TO SEE A SAMPLE TRIAL.

The practice period. At this point the monitor displayed the following:

A B C B A 1 2 3 4 5

USING THE KEYBOARD, PUSHING THE NUMBER UNDER THE LETTER WILL CHANGE IT. TRY IT A FEW TIMES. PUSH # 1,2, 3, 4, AND 5, A FEW TIMES EACH. AND THEN, WHEN YOU ARE READY TO CONTINUE, PUSH THE SPACE BAR.

Pushing Key 1 changed the letter displayed above the 1 from A to B. Continuing to push Key 1 changed the letter displayed from B to C and then back to A. The letters could be recycled indefinitely. Pushing Keys 2, 3, 4, and 5 changed the letters they controlled in a similar manner. Each response also produced a 1-s tone. The following message was displayed after the subject pushed the space bar:

OK, WE ARE READY TO BEGIN. DURING EACH TRIAL, YOU CAN CHANGE THE LETTERS ANY WAY THAT YOU WISH. THE COMPUTER WILL TELL YOU WHEN YOU HAVE CORRECTLY SOLVED EACH PROBLEM BY MAKING CERTAIN LETTERS APPEAR. GOOD LUCK. PUSH THE SPACE BAR TO CONTINUE.

The trial periods. After the subject pressed the space bar, the first of 36 trials began. The stimulus display appeared like this:

B A A A A 1 2 3 4 5

PUSH KEYS 1-5 TO CHANGE LETTERS

36 TRIALS LEFT

The letters were 1.27 cm tall with 5.7 cm between them, center to center. The rule was simple. The subject was to make all five letters the same. In present example, the easiest way to succeed was to push Key 1 twice, thereby producing five As. The subject could also push Keys 2-5, once each (four responses), and produce five Bs. Or the subject could push all five keys to produce five Cs (nine responses).

The reinforcement period. The following screen was presented after the subject made all five letters the same (As in this example):

A A A A A 1 2 3 4 5

YOU ARE CORRECT! GET READY TO CONTINUE

The letters displayed were five As, Bs, or Cs depending on the subject's strategy. During this reinforcement period the background color on the monitor changed from blue to red, and three 1-s tones were presented. This interval was followed by the next trial until the 36 trials were completed. The three letters were presented at trial onset equally often across the 36 trials. Half of the trials presented four identical and one nonmatching letter, for example, AAAAB, CACCC, BBBCB. The other 18 trials presented three matching letters with two identical nonmatching letters, for example, AACCC, BCCBB, CCAAC. Table 1 shows examples of each of the 12 trial types that were presented at trial onset. However, no two trials were identical. Although three trials presented four As and a B, at trial onset, they were presented as BAAAA, AAABA, or ABAAA, once each.

The following display was presented after the 36th trial was completed:

AWESOME! THIS ENDS THE EXPERIMENT EXCEPT FOR A FEW QUESTIONS. PLEASE PUSH THE RETURN KEY TO CONTINUE.

The computer then 'asked the subjects to type in their age, major, grade point, sex, race, American College Test (ACT) score, status (first year, sophomore, junior, or senior), and the rule or strategy that they used. It also asked the subjects to enter any comments that they wished to make. This was followed by the last screen being presented:

THANK YOU AND PLEASE DO NOT DISCUSS THIS EXPERIMENT WITH ANYONE.

After the subject left the room, the experimenter returned, collected the data, and reset the program for the next subject. This condition replicated the Pisacreta (1982b) procedure with four changes. Letters of the alphabet were substituted for colors, the subjects were humans, no grain was presented, and the subjects responded on a keyboard instead of response keys.

Condition 2 - Three Letters: A, B, C, All Present at Trial Onset

Sixteen subjects participated in Condition 2. All the experimental conditions described in Condition 1 were the same except for the stimuli configuration presented at the beginning of each trial. All three letters were present at trial onset. Half of the trials presented two pairs of identical letters and one nonmatching letter, for example, AABBC, CCBAA, CABCA. The other 18 trials presented three matching letters with two nonmatching letters, for example, ABBCB, AACAC, CBCAC. Table 1 shows examples of each of the trial types that were presented at trial onset. This condition was expected to produce more stereotypy than Condition 1 because it would be harder, from trial to trial, to decide which strategy to use, and it would be easier to simply repeat one strategy throughout the session.

Condition 3- Three Letters: A, B, C, Nonrandom Trials

Sixteen subjects participated in Condition 3. All the experimental conditions described in Condition 1 were the same except for the stimuli configuration presented at the beginning of each trial. The same trial types that were employed in Condition 1 were presented, but they were not randomized across the trials. Each subject was first presented with the six trials that presented four As with another letter. These trials were followed by the six trials that presented four Bs at trial onset, and then the four Cs trials (see Table 1). It was not possible to predict whether this condition would produce more or less stereotypy relative to Conditions 1 and 2. Making the letters A six trials in a row could result in the subject generalizing the rule "make them the same" when Trial #7 presented four Bs (less stereotypy). However, the subject having been rewarded for making the letters all A during six trials might perform under the rule "make them all As" during the remaining 30 trials resulting in more stereotypy.

Condition 4 - Three Symbols

Twenty-six subjects participated in Condition 4. The three letters, A, B, C, were directly replaced by three symbols, a horizontal line (-), a cross (+), and a white dot, respectively. These symbols were the same size as the letters used before. Hence, all the stimuli configurations presented at trial onset were the same as those employed in Condition 1, except that symbols substituted for letters. During the practice period, the computer displayed the symbols instead of letters. This condition was expected to produce more or less stereotypy depending upon the subjects' associating more or less significance to symbols relative to letters.

Condition 5- Three Words

Twenty-four subjects participated in Condition 5. The three letters, A, B, C, were directly replaced by three words, NO, MY, and AT, respectively. Hence, all the stimuli configurations presented at trial onset were the same as those employed in Condition 1, except that words substituted for simple letters. These words presumably represented a possessional term (MY), a negative term (NO), and a neutral term (AT). During the practice period, the computer displayed the words instead of letters. This condition was expected to produce more or less stereotypy depending upon the subjects' associating more or less significance to words relative to letters.

Condition 6 - Three Letters: A, B, C, Points

Twenty-five subjects participated in Condition 6. All the experimental conditions described in Condition 1 were the same except for the following. An additional message was presented after the practice period. The monitor displayed:

YOU WILL EARN TEN POINTS EVERY TIME THAT YOU USE THE BEST SOLUTION TO THE PROBLEM. YOU WILL LOSE FIVE POINTS EVERY TIME THAT YOU USE A POOR SOLUTION. TRY TO EARN AS MANY POINTS AS YOU CAN.

WHEN YOU ARE READY TO CONTINUE, PUSH THE SPACE BAR.

An additional message was presented during the reinforcement period. Under the words "YOU ARE CORRECT," the monitor presented either "YOU HAVE EARNED TEN POINTS," or "YOU HAVE LOST FIVE POINTS." Subjects earned 10 points during every trial that they used the most efficient solution. For example, presented with ABBBB, they pressed Key 1 once to replace the A with a B, and thereby end the trial. Subjects lost five points (in the present example) if they used an inefficient solution, that is, they made all the letters A (eight responses) or C (six responses). Each subject's score was displayed in the upper right corner of the monitor during each trial. Providing additional feedback and additional reinforcement (points) was expected to reduce stereotypy.

Results

Figure 1 shows the percentage of subjects who failed to use a parsimonious rule during 75% or more of the trials (solid bars). The shaded bars depict the mean efficiency of subjects' performances. Efficiency was calculated by dividing the minimum number of responses necessary to finish the 36 trials by the mean number of responses made by all subjects who failed to use an efficient rule.

The figure shows that 11 of 24 subjects (46%) in Condition 1 failed to use the rule, "Make all the letters the same with the least number of responses." Similar to the results reported by Pisacreta (1982b) who employed pigeons, the human subjects typically produced a superstitious strategy that yielded response stereotypy during the majority of the trials. When asked the question "What rule did you use?", they answered "Make them all the letter A (or B, or C)," "Make them A then B then C (on consecutive trials)," or "Make them the first letter." Eighty responses were the minimum required to complete the session. Subjects who used a parsimonious strategy typically produced efficiencies that were over 90%. As Figure 1 shows, subjects who failed produced only a 50% efficiency rating. That is, they each make about 160 responses during the session, which is what the pigeons did in the original study.

The second set of bars (Condition 2) shows that having all three letters present at trial onset increased the number of subjects who failed to use a rule to 65%. Efficiency was only 5% higher than that produced in Condition 1.

When the trial types were not randomized (Condition 3), only 5 of 16 subjects (30%) failed to use an efficient rule. However, the efficiencies of those subjects who failed was only 55%.

A comparison of Condition 1 (three letters) with Condition 4 (three symbols) shows that symbols produced less stereotypy than letters, that is, only 8 of 26 subjects (30%) failed in Condition 4 relative to 45% of the subjects in Condition 1. Similar to Condition 1, subjects who failed to use an efficient strategy reported that the rule was "Make them all dots crosses or lines)."

Presenting words as stimuli (Condition 5) produced less failures (20%) and better efficiency than either three letters or three symbols. Superstitious strategies included "Make five MYs appear."

Providing points as feedback yielded the lowest failure rate (5%) of any of the six conditions, presumably because the subjects were punished by point loss if they engaged in stereotyped response sequences and rewarded with points for using a rule.

A one-way analysis of variance (ANOVA) comparing the performances of the subjects across the six conditions of Experiment 1 proved significant, F(5, 126) = 6.64, p [is less than] .001. Efficiency differences across the six conditions were not significant, F(5, 126) = 1.55, p [is less than] .179. There was no significant correlation between performance and: status (first year, sophomore, junior, senior), F(3, 128) = 0.94, p [is less than] .423; or race, F(2, 129) = 0.22, p [is less than] .795. However, males performed better than females, F(1, 130) = 10.89, p [is less than] .001. Several Pearson-Product Moment Correlation tests revealed that the other subject-relevant variables were not reliably correlated with success or failure. There was no significant correlation between performance and: grade point, r(130) = -.05, p [is less than] .557; ACT scores, r(130) = -.01, p [is less than] .823; or age, r(130) = -.09, p [is less than] .255.

There was no significant correlation between efficiency and: status, F(3, 128) = 0.20, p [is less than] .995; or race, F(2, 129) = 0.79, p [is less than] .455. Males performed more efficiently than females, F(1, 130) = 6.59, p [is less than] .011. Comparing the efficiencies and other subject-relevant variables across the six conditions all proved insignificant, that is, grade point, r(130) = .12, p [is less than] .149; ACT scores, r(130) = .07, p [is less than] .406; and age, r(130) = -.01, p [is less than] .908.

EXPERIMENT 2

Based on the results obtained in Experiment 1, the purpose of Experiment 2 was to explore five additional conditions that were expected to yield more or less stereotypy relative to the results obtained in Experiment 1. By definition, using a parsimonious rule should be reinforcing for two reasons. First, the efficient rule would minimize the ratio of effort to rewards. Second, it would shorten session duration. Theoretically, if we increase the cost of engaging in inefficient solutions to problems, the subject should be more likely to experiment with different solutions rather than to perseverate in a strategy that requires a lot of effort, that is, less stereotypy. Two of the conditions of Experiment 2 increased the cost of engaging in inefficient strategies by increasing the number of stimuli available.

Discriminability of the problems and solutions may also affect the emergence of stereotypy. If the number of possible solutions is large, so the parsimonious rule applies to a much broader array of stimulus situations, the subjects may be more likely not to discover the rule, and they may rapidly develop stereotyped behavior as an inefficient but successful solution. Two conditions of Experiment 2 assessed this possibility by increasing the number of possible solutions from three (in. Experiment 1) to 120 in Conditions 2 and 3 of Experiment 2.

Method

Subjects and Apparatus

One hundred and twelve college students agreed to serve as subjects in the experiment. They were individually recruited as they passed by the research lab. Each of the subjects participated in only one of the conditions described below and none of them had participated in Experiment 1. The apparatus was the same described in Experiment 1.

[TABULAR DATA FOR TABLE 2 OMITTED]

Procedure

Five different conditions were run during Experiment 2. They are summarized in Table 2.

Condition 1 - Five Letters: A, B, C, D, E

Twenty-four subjects participated. All the experimental conditions described in Experiment 1, Condition 1 were the same except for the stimuli configuration presented at the beginning of each trial. Each subject was presented with five letters to match, A, B, C, D, E. During the practice period, and each trial, pushing Key 1 changed the letter displayed above the I from A to B. Continuing to push Key 1 changed the letter displayed from B to C to D to E and then back to A. The letters could be recycled indefinitely. Pushing Keys 2, 3, 4, and 5 changed the letters they controlled in a similar manner. Twenty of the trials each presented four identical letters and one nonmatching letter, for example, AAAAC, EBEEE, CCDCC. The five letters (A,B,C,D,E) were the dominant letters during four trials each. The other 15 trials presented three matching letters with two nonmatching letters, for example, ABBCB, DDCDA, EEEAC. The 36th trial presented A,B,C,D,E. Table 2 shows examples of each of the trial types that were presented at trial onset. We believed that increasing the number of stimuli presented during a trial would reduce the emergence of stereotypy. More responses would be required to match the same letter, trial after trial, in this condition relative to Experiment 1, Condition 1. If the subject was presented with BCCCC at trial onset in Experiment 1, and they wanted to make all the letters A, six responses were required because the letters could be cycled only through A,B,C. In the present condition, where the letters cycled through A,B,C,D,E, 16 responses would be required to produce five As.

Condition 2 - Five Letters: No Adjacent

Twenty-four subjects participated in Condition 2. All the experimental conditions described in Condition I were the same. The rule in effect was that no two adjacent letters could be the same. For example, changing the letters AAAAB to ABABC, or EABCE would produce reinforcement. This condition was expected to produce more stereotypy than Condition 1 because it would be harder, from trial to trial, to decide which strategy to use, and it would be easier to simply repeat one strategy throughout the session.

Condition 3 - Five Letters: Make Them Different

Twenty-four subjects participated in Condition 3. All the experimental conditions described in Condition 1 were the same. The rule in effect was that all five letters had to be present in order to complete a trial. For example, changing the letters AAAAB to ABCDE, or EABCD would produce reinforcement. There were 120 solutions possible from trial to trial. This condition was expected to produce more stereotypy than Conditions 1 or 2 because it would be harder, from trial to trial, to decide which strategy to use, and it would appear easier to simply produce one sequence, over and over, throughout the session.

Condition 4 - Five Symbols

Twenty-four subjects participated in Condition 4. The five letters, A,B,C,D,E, were directly replaced by five symbols, a horizontal line (-), a cross (+), a white dot, a spade, and a triangle, respectively. The spade resembled the symbol used on playing cards. These symbols were the same size as the letters used before. Hence, all the stimuli configurations presented at trial onset were the same as those employed in Condition 1, except that symbols substituted for letters. During the practice period, the computer displayed the symbols instead of letters. This condition, like Condition 1 (five letters) was expected to produce less stereotypy. Increasing the number of stimuli presented during a trial was expected to reduce the emergence of stereotypy. More responses would be required to match the same symbol, trial after trial, in this condition relative to Experiment 1, Condition 4 (three symbols). This condition also enabled a direct comparison of three letter, three symbol, five letter, and five symbol conditions.

Condition 5 - Three Stimuli, No Repeats

Sixteen subjects participated in Condition 5. The same conditions presented in Experiment 1, Condition 1 were in effect except that each trial presented three letters, or numbers, for matching. Table 2 shows the stimuli presented at trial onset in the order in which they were presented. Only seven trials presented the same three stimuli but the spatial configuration of the stimuli was never presented before (e.g., Trial 1 = BACAA, Trial 31 = ACBBB). This condition was expected to produce less stereotypy because the subjects could not make all five stimuli the same letter, or number, trial after trial, for example, always make them As. However, the subjects could still engage in stereotyped behavior, for example, make all five stimuli the same as the one on the left.

Results

Figure 2 shows the percentage of subjects who failed to use a parsimonious rule during 75% or more of the trials (solid bars). The shaded bars depict the mean efficiency of the subjects' performances.

As predicted, the figure shows that compared to Experiment 1, Condition 1 (three letters) during Experiment 2, Condition 1 (five letters), only five of the 24 subjects (20%) failed to use the rule, "make all the letters the same with the least number of responses." A t test for a difference between two independent means (three letters vs. five letters) proved significant, t(45) = 5.39, p [is less than] .001. As in Experiment 1, the subjects who failed typically produced a superstitious strategy that yielded response stereotypy during the majority of the trials. When asked the question "What rule did you use?", they answered "Make them the letter A (or B, or C, or D, or E)," Make them A then B then C (on consecutive trials)," or "Make them the first letter." One hundred and twenty-six responses were the minimum required to complete the session. The subjects who used a parsimonious strategy typically produced efficiencies that were over 90%. As Figure 2 shows, subjects who failed produced only a 33% efficiency rating. That is, they each make about 370 responses during the session.

The second set of bars (Condition 2) shows that requiring that no two adjacent letters match increased the number of subjects who failed to use a parsimonious rule to 80%, that is, 19 of 24 subjects. Efficiency was about the same as that produced in Condition 1. Superstitious strategies included "Make them ABCDE (or CBCDA, or some other combination)."

When subjects were required to make each of the five letters appear (Condition 3), 21 of 24 subjects (87%) failed to use an efficient rule, and the efficiencies of those subjects who failed was only 45%. An efficient performance required 221 responses. Subjects who failed typically emitted over 460 responses. Subjects who succeeded produced 25-31 different letter sequences because they changed only the letters that were matches for other letters. Subjects who failed typically used only two to eight sequences because they repeatedly produced ABCDE, or some other combination.

Presenting five symbols in Condition 4 yielded the least amount of stereotypy, that is, 5% (1 of 24 subjects). In Experiment 1, presenting three symbols produced less stereotypy than presenting three letters [ILLUSTRATION FOR FIGURE 1 OMITTED]. Likewise, in Experiment 2, presenting five symbols (Condition 4) yielded less stereotypy than presenting five letters (Condition 1). A comparison of the differences in performances obtained in Experiment 1, Condition 4 (three symbols) and Experiment 2, Condition 4 (five symbols) proved significant, t(48) = 3.56, p [is less than] .001. Similar to Experiment 1, the subjects who failed to use an efficient strategy reported that the rule was "Make them all dots (crosses or lines)."

Surprisingly, presenting nonrepeating letters as stimuli (Condition 5) produced about a 31% failure rate, that is, 5 out of 16 subjects. Although the subjects could not make the letters the same (e.g., all As), trial after trial, a third of them still developed superstitious strategies such as "Make them all the first letter or number." The subjects who succeeded correctly identified the most efficient strategy, for example, "Make them all match the majority present." An efficient strategy required 108 responses. As Figure 2 shows, the subjects who failed emitted over 200 responses.

Statistical analyses were conducted on Conditions 1, 4, and 5 separate from Conditions 2 and 3. Conditions 1, 4, and 5 permitted the subjects to use between three and five solutions to each problem. Conditions 2 (no two adjacent letters could be identical) and 3 (make all five letters different) allowed 120 different solutions to each problem.

An ANOVA comparing the performances of the subjects across Conditions 1, 4, and 5 of Experiment 2 proved significant, F(2, 61) = 5.76, p [is less than] .005. Efficiency differences across the three conditions were also significant, F(2, 61) = 10.80, p [is less than] .001. There was no significant correlation between performance and: status (first year, sophomore, junior, senior), F(3, 60) = 2.66, p [is less than] .056; or race, F(1, 61) = 0.001, p [is less than] .969. The differences in performances between males and females observed in Experiment 1 were not replicated in Experiment 2, F(1, 62) = 2.18, p [is less than] .144. The other subject-relevant variables were not reliably correlated with success or failure. There was no significant correlation between performance and: grade point, r(62) = .01, p [is less than] .925; ACT scores, r(62) = .07, p [is less than] .550; or age, r(62) = -.05, p [is less than] .659.

There was no significant correlation between efficiency and: status, F(3, 60) = 2.64, p [is less than] .057; or race, F(1,61) = 0.000, p [is less than] .976. Contrary to Experiment 1, males did not perform more efficiently than females, F(1, 62) = 3.90, p [is less than] .053. Comparing the efficiencies and other subject-relevant variables across the three conditions all proved insignificant, that is, grade point, r(62) = .09, p [is less than] .435; ACT scores, r(62) = .13, p [is less than] .302; and age, r(62) = -.06, p [is less than] .611.

Comparing the performances of the subjects across Conditions 2 and 3 of Experiment 2 proved significant, F(1, 46) = 39.75, p [is less than] .001. Efficiency differences across the two conditions were also significant, F(1, 46) = 25.39, p [is less than] .001. There was no significant correlation between performance, or efficiency, with status, sex, grade point, ACT scores, race, or age.

GENERAL DISCUSSION

Similar to the results reported by others, many of our subjects produced stereotyped response sequences although the procedures allowed variability. Across the two experiments, 91 of 243 subjects (37%) failed to learn and to use a parsimonious rule. If we ignore Experiment 1, Condition 6 (points earned) and Experiment 2, Condition 4 (five symbols), 89 of 194 subjects (46%) produced response stereotypy.

The main results of the two experiments are as follows. Stereotypy decreased as the number of stimuli available increased, that is, Experiment 1, Conditions 1 and 4 (three letters or three symbols) compared to Experiment 2, Conditions 1 and 4 (five letters or five symbols). Presumably, the additional response cost required to always make five stimuli one particular match, compared to making only three stimuli a particular match, increased the likelihood that the subjects would come under the control of a parsimonious rule. However, increasing the number of solutions available, Experiment 2, Condition 2 (no adjacent matching stimuli) and Condition 3 (make all five letters appear) increased the likelihood that the subjects would emit a single sequence repeatedly.

Having all three letters present at trial onset (Experiment 1, Condition 2) produced more stereotypy than having only two letters present at trial onset (Experiment 1, Condition 1). Presenting several consecutive trials, wherein the simplest solution is the same strategy, reduced stereotypy (Experiment 1, Condition 3). However, a third of the subjects produced stereotypy even under conditions where the stimuli presented at trial onset were seldom repeated (Experiment 2, Condition 5).

The nature of the stimuli also appeared to influence the emergence of stereotypy. Three words (Experiment 1, Condition 5) produced less stereotypy than three letters or three symbols (Conditions 1 & 4). Similar to the results reported by Schwartz (1982), very little stereotypy develops if subjects are reinforced for learning a parsimonious rule, Experiment 1, Condition 6 (points). Curiously, because the points had no monetary value, subjects used a parsimonious rule for no other reason than to earn worthless points.

One purpose of the experiments was to compare human and animal performances on similar tasks. Few would claim that the cognitive abilities of animals are comparable to humans. However, the present data suggest that animals may not be as vacuous as some literature would have us believe. Experiment 1, Condition 1 demonstrated that when humans are trained under conditions comparable to those used with pigeon subjects (Pisacreta, 1982b), almost half of the human subjects fail to behave under the control of a parsimonious rule. Given that humans are capable of learning rules, the failure of pigeons to demonstrate concept learning ability, under certain circumstances, may be the result of the procedures used, rather than a lack of ability on the part of the subjects.

Several papers have suggested that pigeons will learn a concept if it is behaviorally economical to do so (e.g., Fujita, 1983; Herrnstein et al., 1976, Pisacreta, 1996). Some of the best demonstrations of concept learning (rule governed) behavior in nonhumans have involved procedures that employed many stimuli, complex contingencies, and multiple response locations (e.g., Bhatt, Wasserman, Reynolds, & Knauss, 1988; Herrnstein et al., 1976; Pisacreta, Redwood, & Witt, 1984; Pisacreta, Gough, Kramer, & Schultz, 1989; Pisacreta, 1993; Wright, Cook, Rivera, Sands, & Delius, 1988).

In contrast, some of the papers that have reported failure of animals to learn a concept have typically employed procedures that initially trained the subjects to obtain reinforcers by learning a few two-component chains or three-key discriminations (e.g., Carter & Werner, 1978; Cumming & Berryman, 1961; Farthing & Opuda, 1974). The present research shows that even subjects with concept learning ability (humans) will fail to learn a parsimonious rule if they are trained on a procedure that reinforces simple repetition of a response chain. Based on the literature, the failure of humans to come under the control of parsimonious rules appears to be a robust phenomenon, under a variety of conditions (Gross & Gutman, 1988; Heltzer & Vyse, 1994; Lee, 1996; Lopatto & Brown, 1994; Schwartz, 1982). The failure of infrahumans to demonstrate concept learning may be more of a reflection of inadequate training procedures rather than a lack of rudimentary cognitive abilities. More research that trains animals on paradigms originally designed for humans, and vice versa, is required before we can come to definitive conclusions concerning the comparability of animal and human cognition, (see Wasserman, 1997, for an excellent review).

Overall, the data support the position that behavior that is under the control of a superstitious rule typically develops into stereotyped response chains (Heltzer & Vyse, 1994). That is, given multiple ways of solving a problem, many of the subjects emitted the same response sequence over and over, and they indicated that they believed that the strategy that they used was necessary.

One variable that is not quantifiable is the sense of relief that subjects reported and how it affected their behavior. During Trial 1, they responded randomly until they adventitiously used a correct solution. Many of the subjects verbally reported that they felt so relieved to be successful that the idea of trying alternate solutions did not occur to them even though a few trials showed them that alternate solutions would work. For example, during Experiment 1, Condition 1 (three letters), a subject might have been making all the letters C. That is, regardless of the stimuli present at trial onset, they would change the letters, from left to right, to a C. If a trial presented ABBBB, when they tried to change the A to a C, they produced five Bs, and the computer told them that they were correct. Even though they saw that five Bs was also a legitimate solution, and required less effort, they continued on subsequent trials to use the strategy "Make all five letters a C." Other papers have reported this phenomenon, which they have labeled "confirmation bias" (e.g., Mynatt, Doherty, & Tweney, 1977, 1978; Schwartz, 1982). That is, when a solution works, humans are reluctant to test alternatives that might negate their solution as the only legitimate one. They simply repeat whatever works, regardless of its inefficiency.

Other studies have used spatial problems wherein each potential solution is about as efficient as any other (Gross & Gutman, 1988; Heltzer & Vyse, 1994; Lopatto & Brown, 1994; Schwartz, 1982). Wong and Peacock (1986) speculated that stereotypy may reduce physical effort and save time. This may be true in spatial problems such as moving a light from the upper left corner of a matrix to the lower right corner (Heltzer & Vyse, 1994; Lopatto & Brown, 1994; Schwartz, 1982). Pressing two response buttons x times each may require less effort than alternating between them. Our experiments involved visual problems where engaging in stereotypy had a price. The present work demonstrated that even when the cost of engaging in a superstitious strategy is hundreds of additional responses, many of the subjects willingly accept that price.

But why? Schwartz (1982) described what he called "The negative effects of reward." He pointed out that variability (via hypothesis testing) is necessary in order to learn a rule or concept. Reinforcement encourages repetition of a behavioral unit that has been successful, and that unit may reduce the likelihood that the subject will learn a rule or concept (McGraw, 1978; McGraw & McCullers, 1974; Schwartz, 1988). "Contingencies of reinforcement that are effective in creating efficient and successful patterns of behavior in the short term may be counterproductive in the long term as patterns of intelligent hypothesis construction and test are required" (Schwartz, 1988, p. 101). Related to this issue, there have been several criticisms of the education system where rote learning is often stressed (e.g., Bloom, 1987; Kohn, 1993). For two decades, I have been teaching students similar to the subjects employed in the two experiments. They come to class ready to recite information that they have memorized the night before. They can tell me about rods, cones, and color vision theories. However, if I describe how the world might be if we continue to pollute, and ask them how the eye would have to evolve in order to still serve us well, a silence falls over the room. Furthermore, many students have told me that they seldom had to apply knowledge in their high school and college courses. Most of the exams that they have taken used a multiple choice format. They are uncomfortable with essay tests and have tried to avoid courses that use them. Neither race, age, sex, or other subject-relevant variables could explain the results reported here, that is, why some subjects used a parsimonious rule whereas others did not. Perhaps, the answer lies in the educational history of the subjects. Those students who were reinforced for developing concepts and engaging in hypothesis testing in the past were more likely to discover and use an efficient rule during the experiments. The other students who may have had a long history of success by rote repetition may have been content to be successful, regardless of how inefficient their strategy was.

In a recent review paper, Eisenberger and Cameron (1996) pointed out that reinforcement may produce or impede task interest and creativity, depending on the procedure employed. That is, whether we produce more or less stereotypy, creativity, and intrinsic motivation depends on the conditions under which we learn. Deci and Ryan (1985) identified three types of reward situations. Performance-independent rewards situations provide rewards for mere participation. Completion-dependent rewards require the subject to complete a given task. Quality-dependent rewards compares the subject's performance to others, or a standard. Eisenberger and Cameron (1996) claimed that "the reinforcement situations most certain to produce detrimental effects on intrinsic interest involve performance-independent rewards and completion-dependent rewards" (p. 1155). The paradigm used in the present research was a completion-dependent rewards situation, and therefore would be expected to yield little creativity.

Reward can strengthen divergent thinking, that is, the development of many creative solutions to the same problem, if the situation is structured to produce it (Goetz & Baer, 1973; Maltzman, 1960; Winston & Baker, 1985). Learned industriousness theory predicts that rewarding a minimum effort will reduce tendencies to engage in more efficient strategy use (Eisenberger, 1992). Humans will learn what leads to reward and adjust their behavior accordingly. If a situation reinforces divergent thinking, it will emerge. However, if the situation will reward rote memorization, or repetition of an inefficient strategy (as the experiments reported here did), that behavior pattern will dominate the subject's performance. We agree with Eisenberger and Cameron (1996) that more research is needed, in a spirit of cooperation among various fields, in order to identify the environmental variables that influence human learning. Like Wasserman (1997), we would like to see those efforts extended to include animal cognition.

In summary, superstitious beliefs and response stereotypy impede the development of efficient rule-governed behavior in humans and nonhumans. Stereotypy may mask concept learning ability. In education, a reinforcement history of rote learning may have a deleterious effect on the development and use of inductive reasoning by students, particularly in situations that neither expect, nor require, the use of divergent thinking.

References

BHATT, R. S., WASSERMAN, E. A., REYNOLDS, W. F., & KNAUSS, K. S. (1988). Conceptual behavior in pigeons: Categorization of both familiar and novel samples from four classes of natural and artificial stimuli. Journal of Experimental Psychology: Animal Behavior Process. 14(3), 219-234.

BLOOM, A. (1987). The closing of the American mind. New York: Simon and Schuster.

CARTER, D. E., & WERNER, D. J. (1978). Complex learning and information processing by pigeons: A critical analysis. Journal of the Experimental Analysis of Behavior, 29, 565-601.

CATANIA, A. C., & CUTTS, D. (1963). Experimental control of superstitious responding in humans. Journal of the Experimental Analysis of Behavior, 6, 203-208.

CUMMING, W. W., & BERRYMAN, R. (1961). Some data on matching behavior in the pigeon. Journal of the Experimental Analysis of Behavior, 4, 281284.

DECI, E. L., & RYAN, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York: Plenum Press.

EISENBERGER, R. (1992). Learned industriousness. Psychological Review, 99, 248-267.

EISENBERGER, R., & CAMERON, J. (1996). Detrimental effects of reward. American Psychologist, 51, 1153-1166.

FARTHING, G. W., & OPUDA, M. J. (1974). Transfer of matching-to-sample in pigeons. Journal of the Experimental Analysis of Behavior, 21, 199-213.

FUJITA, K. (1983). Acquisition and transfer of a higher-order conditional discrimination performance in the Japanese monkey. Japanese Psychological Research, 25, 1-8.

GOETZ, E. M., & BAER, D. M. (1973). Social control of form diversity and the emergence of new forms in children's blockbuilding. Journal of Applied Behavior Analysis, 6, 209-217.

GROSS, D. K., & GUTTMAN, A (1988). Effects of schedule and length of training on reward-induced stereotypy and negative transfer in humans. The Psychological Record, 38, 567-594.

HELTZER, R. A., & VYSE, S. A. (1994). Intermittent consequences and problem solving: The experimental control of superstitious beliefs. The Psychological Record, 44, 155-169.

HERRNSTEIN, R. J., LOVELAND, D. H., & CABLE, C. (1976). Natural concepts in the pigeon. Journal of Experimental Psychology: Animal Behavior Processes, 2, 285-302.

HULSE, S. H., FOWLER, H., & HONIG, W. K. (Eds.). (1978). Cognitive processes in animal behavior. Hillsdale NJ: Erlbaum.

KELLER, F. S., & SCHOENFELD, W. N. (1950). Principles of psychology New York: Appleton-Century-Crofts.

KENDRICK, D. F., RILLING, M. E., & DENNY, M. R. (Eds.). (1986). Theories of animal memory Hillsdale NJ: Erlbaum.

KOHN, A. (1993). Punished by rewards. Boston: Houghton Mifflin.

LEANDER, J. D., LIPPMAN, L. G., & MEYER, M. E. (1968). Fixed interval performance as related to subjects' verbalizations of the reinforcement contingency. The Psychological Record, 18, 469-474.

LEE, V. L. (1996). Superstitious location changes by human beings. The Psychological Record, 46, 71-86.

LOPATTO, D. E., & BROWN, B. M. (1994). The effects of schedules of reinforcement on operant sequential stereotypy in humans. The Psychological Record, 44, 185-206.

MALTZMAN, L. (1960). On the training of originality. Psychological Review, 67, 229-242.

McGRAW, K. O. (1978). The detrimental effects of reward on performance. A literature review and a prediction model. In M. R. Lepper & D. Greene (Eds.), The hidden costs of reward (pp. 33-60). Hillsdale, NJ: Erlbaum.

McGRAW, K. O., & McCULLERS, J. C. (1974). The distracting effect of material reward: An alternate explanation for the superior performance of reward groups in probability learning. Journal of Experimental Child Psychology, 18, 149-158.

MYNATT, C. R., DOHERTY, M. E., & TWENEY, R. D. (1977). Confirmation bias in a simulated research environment: An experimental study of scientific inference. Quarterly Journal of Experimental Psychology, 29, 85-95.

MYNATT, C. R., DOHERTY, M. E., & TWENEY, R. D. (1978). Consequences of confirmation and disconfirmation in a simulated research environment. Quarterly Journal of Experimental Psychology, 30, 395-406.

PISACRETA, R. (1982a). Some factors that influence the acquisition of complex stereotyped response sequences in pigeons. Journal of the Experimental Analysis of Behavior, 37, 359-369.

PISACRETA, R. (1982b). Preferences among stimulus matches in the pigeon. Journal of the Experimental Analysis of Behavior, 38, 191-199.

PISACRETA, R. (1990). Rudimentary rule-governed behavior in the pigeon. In M. L. Commons, S. M. Kosslyn, R. J. Herrnstein, & D. G. Mumford (Eds.), Quantitative analyses of behavior, Vol. 8: Behavioral approaches to pattern recognition and concept formation (pp. 23-39). Hillsdale, NJ: Erlbaum.

PISACRETA, R. (1993). Establishment and transfer of multiple conditional discriminations in the pigeon. The Psychological Record, 43, 59-84.

PISACRETA, R. (1996). Transfer of oddity-from-compound samples in the pigeon: Some assembly required. Behavioral Processes, 37, 103-124.

PISACRETA, R., GOUGH, D., KRAMER, J., & SCHULTZ, W. (1989). Some factors that influence transfer of oddity performance in the pigeon. The Psychological Record, 39, 221-246.

PISACRETA, R., REDWOOD, T., & WITT, K. (1984). Transfer of matching to figure samples in the pigeon. Journal of the Experimental Analysis of Behavior, 42, 223-237.

PISACRETA, R., REDWOOD, T., & WITT, K. (1987). Stimulus control of the pigeon's ability to peck multiple relocating targets: Stereotypy prevents effective strategies. The Psychological Record, 37, 563-585.

PREMACK, D. (1978). On the abstractness of human concepts: Why it would be difficult to talk to a pigeon. In S. H. Hulse, H. Fowler, & W. K. Honig (Eds.), Cognitive processes in animal behavior (pp. 423-451). Hillsdale, NJ: Erlbaum.

ROITBLAT, H. L., BEVER, T. G., & TERRACE, H. S. (Eds.). (1984). Animal Cognition. Hillsdale, NJ: Erlbaum.

SCHWARTZ, B. (1980). Development of complex, stereotyped behavior in pigeons. Journal of the Experimental Analysis of Behavior, 33, 153-166.

SCHWARTZ, B. (1982). Reinforcement-induced response stereotypy: How not to teach people to discover rules. Journal of Experimental Psychology: General, 111,23-59.

SCHWARTZ, B. (1988). The experimental synthesis of behavior. Reinforcement, behavioral stereotypy, and problem solving. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 22, pp. 93-138). San Diego: Academic.

SKINNER, B. F. (1948). "Superstition" in the pigeon. Journal of Experimental Psychology, 38, 168-172.

SLAVIN, R. E. (1997). Educational psychology: Theory and practice (5th ed.). Boston: Allyn and Bacon.

SPEAR, N. E., & MILLER, R. R. (Eds.). (1981). Information processing in animals: Memory mechanisms. Hillsdale, NJ: Erlbaum.

VOGEL, A., & ANNAU, Z. (1973). An operant discrimination task allowing variability of response patterning. Journal of the Experimental Analysis of Behavior, 20, 1-6.

WASSERMAN, E. A. (1997). The science of animal cognition: Past, present, and future. Journal of Experimental Psychology: Animal Behavior Processes, 23, 123-135.

WASSERMAN, E. A., KIEDINGER, R. E., & BHATT, R. S. (1988). Conceptual behavior in pigeons: Categories, subcategories, and pseudocategories. Journal of Experimental Psychology: Animal Behavior Processes, 14, 235-246.

WONG, P. T., & PEACOCK, E. J. (1986). When does reinforcement induce stereotypy? A test of the differential reinforcement hypothesis. Learning and Motivation, 17, 139-161.

WINSTON, A. S., & BAKER, J. E. (1985). Behavior analytic studies of creativity: A critical review. Behavior Analyst, 8, 191-205.

WRIGHT, A. A., COOK, R. G., RIVERA, J. J., SANDS, S. F., & DELIUS, J. D. (1988). Concept learning by pigeons: Matching-to-sample with trial unique picture stimuli. Animal Learning and Behavior, 16, 436-444.
COPYRIGHT 1998 The Psychological Record
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Pisacreta, Richard
Publication:The Psychological Record
Date:Mar 22, 1998
Words:9051
Previous Article:Effects of response cost in computerized programmed instruction.
Next Article:Effects of differing instructional histories on the resurgence of rule-following.
Topics:


Related Articles
Intermittent consequences and problem solving: the experimental control of "superstitious" beliefs.
The effects of schedules of reinforcement on operant sequential stereotypy in humans.
The analysis of rule-governed behavior in social psychology.
Superstitious math performance: interactions between rules and scheduled contingencies.
Effects of differing instructional histories on the resurgence of rule-following.
Use of self-monitoring and delayed feedback to increase on-task behavior in a post-institutionalized child within regular classroom settings.
Toward a socially valid understanding of problem behavior.
Emerging Equine Science.
Schedule-induced and operant mechanisms that influence response variability: a review and implications for future investigations.
The effects of a stimulus-stimulus pairing procedure on the acquisition of conditioned reinforcement on observing and manipulating stimuli by young...

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |