Intro
Operant conditioning is a style of conditioning that is characterized by the giving or taking of punishment and reinforcements in an attempt to teach a person a desired behavior (Skinner, 1963). This sort of conditioning is commonly used with animals to coerce them into completing a desired behavior, such as to sit or lay down. Operant conditioning of this sort should have an acceptable reinforcing stimulus in order to enhance the frequency of the desire behavior (Greenspoon, 1955). Previous research tells us that verbal reinforcers such as “mmmhm” or “good” are effective as intrinsic motivators (Verplanck & Wilson, 1956). Additionally, the clicker is used more commonly with animals, in which a trainer might say a command or wait for the animal to do a desired behavior, then will click the clicker, and the animal will receive a treat. Food is used as an intrinsic motivator for the animals to do the desired behavior, but the clicker is used as a predecessor to the food so that the animal will view the clicker as part of the reinforcement (Pryor, 1999). After this pairing is completed the clicker will be seen as a primary reinforcer, as opposed to the secondary reinforcer when paired with a food reinforcement (Pryor, 1999). An example of clicker training put into practice was in the Rotenberg (2015) study, where researchers gathered 27 dogs and tried to condition a simple trick. The trick was for the dogs to stick their front paws into a hoop when the researcher gave a verbal cue. Rotenberg (2015), divided the dogs into two groups, one of which was the NRM group and the other, the control group. The control group was assigned to ignore the undesirable behaviors of the dogs. The NRM group failures were communicated to the dog in a tone that would signal the dog that they made a mistake (Rotenberg, 2015). Rotenberg (2015), found that the control group dogs had better test performance and concluded that dogs learn better when they are unaware they are learning and aren’t stressed about results of conditioning.
Another experiment that investigated clicker training and participant awareness was conducted by Smith and Davis (2008). In this experiment, researchers assigned 35 dogs to touch a cone with their noses and split animals into two groups to conduct three trials for each. The first trial was accompanied by a reinforcement, either a clicker plus food or with food alone, when the animal was training to complete the desired behavior. The second trial was attempting to enhance that learned behavior and reinforcements were given randomly. The final trial was extinction, where food was withheld but the researcher still used the clicker with the group that had been receiving the clicker. Smith and Davis (2008), found no difference between the two groups but found that the dogs that received the click and food for reinforcement had a better procurement of the skill. Many animal trainers and owners use the clicker method and notice a decline in the desired behavior. According to Smith and Ellen (2007), learning will decrease when a behavior is performed or appears and there is no reinforcement following.
With this information, it could be inferred that humans would react the same way. Obviously, humans are not dogs but results from learning studies with dogs can help improve verbal conditioning for humans. To get a better understanding of human learning and impression from the experiment, there should be direct questions to assess their disposition. Gosling, Rentfrow, and Swann (2003), propose that survey questions are the best way to analyze the participant’s attitudes on a specific personality trait. These questions should be transparent and asked directly to get the best results. Ideally, questions would be centered around the big five personality traits: openness to experience, neuroticism, extraversion, conscientiousness, and agreeableness (Gosling, Rentfrow, and Swann, 2003). These personality traits may have a significance on receptivity of condition or rate of learning.
Similar to studies with dogs, humans may also be able to be verbally conditioned the same when they are unaware of the intent. In 1962, Spielberger, Levin, and Shepard found there is an association between the researcher’s response and the participants behavior. Although they did not find that the participants unawareness led to higher scores in humans, other research suggests otherwise and prompts us for further investigation.
Method
Participants
Participants for this particular study were selected based on several pieces of criteria. This criterion included that the individual be between the ages of 19-65 and have a good understanding of the English language. There were 36 participants of varying ages, ethnicities, and education levels to help ensure an adequate sampling size.
Experimental Procedure
The first task in this experiment is to assign reinforcement conditions (Clicker, Yes/No, and Yes) randomly to participants and record the participants condition and participant number. Next, the script, found in Appendix A, is read. When the two minutes on the timer starts, each response must be recorded and reinforced in regard to being correct or incorrect and the percentage of correct words is calculated. See Appendix B for recording sheet and survey questions. After this step is repeated four times, the first set of survey questions must be asked and recorded. These questions are important because their purpose is to remove the negative effects of stress of learning (Rotenburg, 2015). These survey questions are given before and after the extinction trial. During the extinction trial, there are no reinforcements or verbal communication given from the researcher. Once the survey questions have been asked again after the extinction trial, the purpose of the experiment is revealed, and more participant background information is recorded. Information such as age, gender, and occupation may have a significant influence on the results. Finally, the personality assessment is given, and TIPI Scores are calculated. Refer to Appendix C for additional survey questions and background information form.
Results Section
In the following study we wanted to see if there would be a change in verbal behavior over the five trials. We used a paired t-test to compare trial one and trial five. We found that there is a significant difference between the two trials, t (36) = -2.281, p =.029. We used a repeated measures ANOVA and we found that there was no main effect for condition, F(2, 33) = 1.909, p = .164. We did find however there was a change over time, F(5,165) = 2.77, p = .019. Additionally, we did see a change across trials 1-5 plus extinction. There was no interaction. See appendix D for reference.
We then wanted to see if the extinction trial would lead to fewer plurals compared to Trial 5 and so we used a paired t-test between trial 5 and the extinction trial. We found there was marginal significance between the two, t(36) = 1.988, p = .054. We used a repeated measures ANOVA and found there was no significant for the extinction trial across conditions as well, F(2,34) = 2.2, p = .126. See Appendix E for reference.
In wanting to see if each condition would show a significant decrease in plurals from trial 5 to extinction, we conducted a paired t test for each condition. We found that it was significant in the condition of the clicker, t(12) = 2.3, p = 0.04. We also found that it was no significant in the Yes trial, t(10) = .539, p = 0.04. Finally, we saw found that the Yes/No trial was no significant as well, t (12) = -.714, p = .489.
We found no effects of gender, occupation, previous clicker training experience, or age in percentage of correct plurals and extinction trials. We wanted to know if the participations did report enjoyment and recommendation of the activity change over time and amongst conditions. We conducted a repeated measures ANOVA and found that there was no significant effect, F(2,24) = .338, p = .716. We wanted to know if there was an effect of gender on the amount of enjoyment of the experiment as well. After performing a repeated measures ANOVA, we found that was an interaction but no main effect across trials, F(2, 25) = 4.62, p = .014. Another question we had was whether the condition would have a significant effect on the recommendation of the participant to another person and conducted a repeated measures ANOVA. We found there was no main effect for condition or change over time, F(2,48) = 32.7, p = <.0001. These results led us to questioning whether there was an effect of gender on the participants recommendation of the experiment and completed a repeated measures ANOVA. We found that there was a main effect, a change over time and an interaction, F(2,25) = 4.67, p = .013. This means that the participants gender impacted their scores over time and had an effect on the value of the score the individual gave.
Finally, we wanted to know if openness effected the learning of the participants and completed a simple regression. We found that there was no effect of openness on learning, R^2(37) = .119, p = .036.
Discussion
This experiment found that there is an effect of reinforcement on unconscious learned behaviors, amongst other findings related to extinction trials and the effects of gender. Our first finding was that participants did increase their use of plurals over time as compared to trial one to trial five. After completing the paired t test and finding that the p value of .028 was lower than p = .05. This could be explained by gradual learning over the trials. The learning would be more evident in the later trials compared to the first trial because learning is better observed when the participant is unaware of the intent of the experiment, and that extinction of this behavior would take longer and requires more trials (Smith & Davis, 2008). Some questions that could be explored based on this finding could be about what would happen with the learning if there were more trials and how would the learning be affected by possible fatigue. Another question proposed was whether the condition was influenced the verbal conditioning and we found that there was no main effect for condition or interaction, but we did see a change across the trials including the extinction trial.
We wanted to find out if the extinction trial would lead to fewer plurals over trial 5 and found that there was marginal significance. This could be explained by (ARTICLE EXPLAINING EXTINCTION). We also wanted to find if there was a difference between conditions in the extinction trial and found that there was no significance because after completing a repeated measures ANOVA the p value was greater than .05. However, there was evidence of significant decrease in plurals from trial 5 to extinction for the clicker condition. There were no gender, age, occupation, or clicker training experience on the percentage of correct plurals and extinction trials.
The survey questions yielded useful results as well. Although there was no significant effect of reported enjoyment and recommendation of the activity between conditions, there was a gender effect. Males and females reported similar results for enjoyment after the two initial surveys were administered but males reported higher enjoyment at the end of the experiment over females. See Appendix F for reference. The same question about effect of condition on recommendation and although there was no main effect, there was a change over time and more specifically in the final set of survey questions. Once again there was a main effect, change over time, and an interaction between genders and their recommendation of the experiment. See appendix G for reference. In contrast to the previous results centered about gender and enjoyment, the men recommended the experiment on average about 2-2.5 points more than the females. Although we did find that both genders’ scores decreased significantly once the final survey questions were asked, with scores dropping as much as 5.5 points. An expansion of this experiment could be to try to have volunteers that may not define themselves by traditional gender roles, such as transgender, males who identify as female and so on. Information like that might change the results because sexual and gender identity could be different or might lead to further investigation of whether females or males enjoy other tasks more or less.
Finally, there was a correlation between openness and the extinction trial. This could be explained by (INSERT ARTICLE ABOUT OPPENNESS)
Conclusion
Per our results, we found there was effective learning when the participant was unaware of the intent of the experiment and that several factors affect the learning and extinction of that learned verbal behavior. We found that participants learned to change their verbal behavior with varying reinforcements, but there is no effect to the type of stimulus on the learning. We also found there is not an impactful difference between the final trial and extinction, even across the conditions. Through analysis of the results we can also conclude that gender has some effect on the enjoyment and recommendation the participant reported. These findings can help better understand how to teach a desirable behaviors to an unknowing subject and could help an educator/trainer better reinforce learning in the classroom and elsewhere. These finding can also help clinical psychologists with verbal conditioning and improving doctor-patient relationship involving therapy or interviews (Kanfer & Karas, 1959).