Peak Shift in Honey Bee, Research Paper Example

Introduction

Stimulus generalization is conventionally defined as conditioning a stimulus to mimic similar responses after the initial response has been conditioned. Stimulus generalization is generally studied by providing images or objects and individuals and then measuring their ability to respond to them, either in terms of speed or consistency of association. Discrimination training conditions people to certain responses and can therefore cause them to react towards certain stimuli in a different manner than expected. This can cause a peak shift effect, and many response variables have the opportunity to impact this phenomenon. Previous research has indicated that the extent to which study participants are provided with discrimination training affects the degree to which a shift occurs (Derenne et al., 2015). Other research has indicated that most animals display decremental generalization gradients in which a peal of responding occurs “at or near the stimulus correlated with reinforcement(S+) and progressively fewer responses occur at stimulus values farther from S+” (Howard, 1979).

Based on the existing research, it is plausible that participants will preferentially remember faces that have exaggerated traits, which are indicative of their sexual preferences (ten Cate et al., 2006). It is therefore important to gain an understanding of whether this effect will hold for individuals that receive discrimination training versus those who do not. Typically, participants are trained with a single S+ and single S-, but this practice does not establish external validity. Therefore, it is necessary to gain a better discernment of this relationship. It is hypothesized that the peak will shift away from the S-. If the participants are trained to not respond to a neutral face, then they will disproportionately respond to the fearful faces due to an overcompensation. It is important to consider that while this situation is relying upon the visual sense, the other four senses have the ability to impact this relationship in nature (Andrew et al., 2013). Therefore, this analysis is assessing this relationship without these confounding variables being present in the study.

Method

Participants

A total of 17 participants were recruited to the study to test condition 1 (S- is fearful and one S+), 22 participants were assigned to test condition 2 (S- is neutral and a range of S+s), 15 participants were assigned to test condition 3 (S- is neutral and one S+), and 8 participants were assigned to test condition 4 (S- is fearful and a range of S+s).

Apparatus

Nine faces ranging from a neutral expression to a fearful expression were collected. Each face was designed to appear six times. The S+ was indicative of a face that was half way between a neutral and fearful expressions. The S- was indicative of a neutral expression or a fearful face. One half of the participants were trained with one of the faces being the S+ and one of the faces being a S-, while the other half of the participants were trained with three different intermediate faces (face 4,5,6) as S+’s and 3 other faces (1,2,3) as S-. A computer program displaying the faces were provided to the participants. Participants received discrimination training according to one of the four conditions prior to the beginning of the experiment. After this period, they were then provided with a generalization test.

Procedure

2×2 factorial design was used and the selection of the S- was treated as an independent variable. All of the participants received discrimination training; half were trained to not respond to the neutral face and the other half were trained to not respond to a fearful face. The second independent variable involved the number of stimuli that were included in the discrimination training. Over the course of the procedure, participants received 30 discrimination trials and then proceeded to the generalization test of 54 trials. All face identification procedures occurred in front of a computer screen. During the training process, the researcher indicated which selection was correct and which was not; this feedback was not provided during the trial. SPSS was used for statistical analysis.

Results

A descriptive statistics analysis of the data indicated that the condition 1 group incurred an average of 1.34 error, condition 2 incurred an average of 2.53 errors, condition 3 incurred 0.60 errors, and condition 4 incurred 3.12 errors. Thus, it is evident that condition 3 experienced the highest level accuracy, while condition experienced the least. Overall this indicates that it, it was more difficult to discriminate one a fearful S- than on a neutral S-. An ANOVA was utilized to determine whether the results received were due to chance, or whether they were statistically meaningful. It was found that the neg. stim value was .000, which is a statistically significant value. For the purposes of this test, all values below 0.05 were held to be statistically significant. It was found that the independent variable number stim was .000, which is also a statistically significant result. On the other hand, it was found that the interaction between the values was not significant.

Discussion

Since the ANOVA test yielded a p-value of less than 0.05 for both relationships, it is apparent that a relationship was found to exist between the different study groups. Overall, the hypothesis was correct because the peak did shift away from the S-. In addition, it was found that the training received by each group helped them recognize a different range of faces. This shows that when we think about people or faces in a certain way, we tend to continue to perceive them in this manner. Although this experiment was conducted in a controlled environment, this process is reflective of the way that we recognize faces and objects in our own environment. When we become accustomed to thinking of certain things in specified ways, it becomes challenging to think about them in a different manner. It is therefore important to understand that these experimental results were received in part due to the way that humans are trained to perceive and remember certain images.

It is also interesting to consider the confounding variables that are present in this relationship. First, it is challenging to determine whether these results would be externally valid due to the small sample size. Second, many characteristics impact the ability for an individual to recognize a face. In fact, individuals with schizophrenia have very poor facial recognition. Although it is unlikely due to probability that the individuals that participated in this experiment had biased results due to the existence of this disorder, it is necessary to consider some of the confounding variables that exist in the facial recognition and perception relationship for future experiments to ensure that the effect that is being studied is truly reflective of the results, instead of being indicative of some other phenomenon.

A greater degree of statistical analysis would also allow these confounding variables to be removed from the analysis. Instead of training each individual, participants could be matched to controls that have similar demographics and health characteristics as them. It would be interesting to compare this case-control study to other study formats to determine which would produce the most accurate results. Overall, many different variations of this study are possible.

References

Andrew SC, Perry CJ, Barron AB, Berthon K, Peralta V, Cheng K. (2013). Peak shift in honey bee olfactory learning. Anim Cogn, DOI 10.1007/s10071-014-0750-3.

Derenne A, Loshek EA, Bohrer B. (2015). Postdiscrimination Gradients With Familiar and Unfamiliar Faces. Psychol Rec, 65: 77–82.

Howard RW. (1978). Stimulus Generalization Along a Dimension Based on a Verbal Concept. Journal of the Experimental Analysis of Behavior, 32: 199-212.

ten Cate C, Verzijden MN, Etman E. (2006). Sexual Imprinting Can Induce Sexual Preferences for Exaggerated Parental Traits. Current Biology, 16: 1128–1132.