Rule-Based and Association Based Learning Processes - Literature review Example

Rule-Based and Association Based Learning Processes First, Introduction Jones and McLaren in their article of rules and association investigated two learning theories of human cognition. They say that learning can occur through rule-based or associative process. Two experiments were six stimuli which varied along a luminance were done. In the two experiments, the time of learning between the participants was varied. In experiment 1, short duration trained participants showed a peak shift curve which was explained as an associative process. Long duration trained participants showed a consistent behavior of rule-based process. In experiment 2, full contingency participants showed rule-based learning while reduced contingency showed an associative process. Keywords; associative process, rule-based process. Rule-Based and Association Learning Processes Methodology The experiment was to run on a computer in a very quiet room with only a small lamp on the desk. The participants in the experiment were divided into two equal numbers representing each condition of the experiment. They were expected to key in the correct stimuli into the computer as soon as the stimuli appear. The first phase was expected to be a training phase, so the computer gives feedback on whether the answer is right or wrong. After the second phase of the same procedure, the participants were required to answer a structured questionnaire. The second phase was a similar procedure, but no feedbacks were given ( Laren, 1994). The Design Used The design used in the first experiment by According to McLaren, Green & Mackintosh (1994) involved a short and long training which took 48 and 96 hours respectively. A pseudo-random sequence was used for the short training with a set of 12 trials while a random sequence was used for the long training with a set of 24 trials. A test filler was equalized for both trainings, and it appeared on odd trials only. The experiment was done in 3 phases that is the training phase; test phase and finally structured questionnaires were administered. In both the training and test phase, the participants were given a computer place one meter away and on eye level. There was an experimenter who gave instructions and left the room after instructions on every phase. The test phase and training phase were similar except for the lack of feedback on their accuracy and the speed of determining the stimuli was higher in the test phase than the training phase. In experiment 2 by McLaren, Green & Mackintosh, two groups were involved. The full contingency which was similar to the long training in the first experiment and the reduced contingency group. The reduced contingency group differed from the short training in that the green stimuli and their respective categories were reduced, and the key assignments were reversed. Eighty participants who were paid were used. Results In experiment 1, it was not surprising that the long based training had more accurate results than short based training. A peak shift is seen for the short training while an increasing trend was seen in long group training. In experiment 2, data was analyzed the same way data in experiment one was analyzed. The full contingency was seen to respond more accurately than the reduced contingency. Reliability was assessed by a planned contrast which showed that the two conditions were significantly different. Analysis The results of experiment 1 from the long training and the short training were distinguished by the near and short stimuli because the two conditions of distance were indeed significant. Analysis of the data again using planned contrast showed that; there was a significant response to near stimuli than to the distant and training stimuli. This suggests that the peak shift was reliable in short training. The presence of peaks in short training and changing gradients were associated with the participant’s incapability to characterize the different stimuli verbally. Consequently, in long training where the participants were able to verbalize the number of underlying rules and stimuli’s. In experiment two, a rule-based process was seen in the full contingency group of participants. The near stimuli participants were not responding accurately unlike the distant stimuli participant who showed a significant accurate response. From the results of full contingency participants, we can conclude that the long training participants were rule-based learners since the two groups were a replication of each other. This result is similar to McLaren, Green & Mackintosh (1994) argue that the increase in the stimuli contingencies complexity increases the amount of information stored in our memory thus fulfilling the rule-based process. Reduced contingency group of participants showed more consistency performance. The one-tailed significance showed peak shifts in this group though they were not reliable. Thus, an associative based performance was depicted in this group of participants From the structured questionnaire, Participants were according to “rule learned” their ability accurately to verbalize the difference that arose between the different stimuli in terms of at least one of the color components and if they were able to make their judgments during the test phase according to color popping up on their screen. This was to determine if, they acquired the rule during training or immediately after, training. All the participants were informed of the dimensional variation post-test. Generally, associative learning and rule-based process can be distinguished by how individuals respond differently to stimuli. The associative process is indicated by a peak shift while the rule-based learning will be indicated by an increasing trend. Both short training and reduced contingency training will always give an associative performance while long training and full contingency will always give a rule-based performance. This can be explained that: with reduced contingency or little training, the relationship will be weak thus a poor rule based performance. The strength of the association increases with further training enabling the rule-based process. (Laren, 1994). These results can be supported by previous works by Aitken, which suggest that more and advanced training should produce a peak shift. (Aitken, in preparation). Critique First an assumption could be made that, the right answers given to the stimuli appearing on the computer screen could be a guess. As opposed to the fact that the results were gotten from the knowledge gained during the training. Participants in the study could actually not be interested in the study thus all work done could be merely guessed work. This could interfere with the experimenter results and conclusion without their knowledge. Secondly learning is not always assumed to begin from associative process rather it is a complex interaction of both associative and cognitive process. Participants who represent all those initiated in the learning process possess very different abilities. Some can start their learning process from simple to complex process while some from complex to simple process. Thus following the conclusion from the experiments, one can be misguided to think all human brains are similar.(Shanks,1995) Another assumption that could be made is, and both patterns i.e., associative and rule-based could be considered as rule-based performance if a complex account or process could be used to determine the formation of a peak shift. For instance, the difference in distance could bring out different results. Equalizing the distance or making both distances longer could lead to a similar performance (Wills $ Mackintosh,1998 Pg 1-1 ) The likelihood of classifying wrongly could be a major difficulty. Those classified as rule-based learners could have fallen in the category of peak shift if extreme stimuli had been used. Conclusion 1 In experiment 1, the short time training participants showed an associative performance. With lengthy exposure, the participants showed rule-based process. In summary experiment, 2 shows that, Full Contingency participants produced a rule-based performance, majority of participants were able to verbalize a rule. The reduced Contingency showed a peak shift pattern. Moreover, only a few participants were able to verbalize the rule. This is similar to McLaren, Green & Mackintosh’s guess that decreasing contingency reduces the likelihood of occurrence of the rule of abstraction. Associative learning thus dominates performance. Implication to Human Beings If there is a peak shift drawn when any difficult perceptual discrimination is both learned and not under the control of some rule, then this could have important implications for human decision making and choice preference. Studying animal learning will help us understand human behavior better (Levesey, 2009,554-565). Further research The need to further classify cognitive and associative process and their interaction. This should be done using a greater number of people who are a proper representation of the larger population. Further analysis In his journal of animal behavior process, Mclaren did two experiments. Participants used were supposed to learn how to differentiate between a pair of similar colors over a range of illuminations. They found out that the gradients of the curves formed a peak shift initially and later an increasing monotonic curve. The presence of the peak performance was related to the participants inability to characterize verbally the difference between the training stimuli,. From these observations, we can derive the conclusion that the transition from using simple physical similarity to verbalize rule is as a result of more information being available when judgment is being made (Livesey,2009 , pg 554-565). Contingency theory formalizes associative learning (Rescorla, 1967). In Rescorla`s "American" view of Pavlovan, conditioning focused on the frequency of pairings between reinforcement and the conditioned stimulus (CS). His approach emphasized exclusively the ability of the conditioned stimuli to be able to send signals to an imminent unconditional signal. From this theory, learning is associated with two conditional probabilities that is the probability of unconditional signal when the conditional signal is given and the probability of the unconditional signal when the conditional signal is absent. Shank in his article also argues that the contingency theory is a summation of the levels of associative learning. (1995) Another study by Blaisdell of the university of California showed that, associative process build on the structure on which cognitive process can act upon. For instance, associative process are involved in causal and special maps. These conclusions are also supported by studies such as pavlovian and sensory preconditioning. A study by Donna which examined the rate at which individuals forget used children in their experiment. From this experiment, it was found that individuals rate of forgetting depends on the working memory performance. The rate at which a child loses memory is a major constraint if the factor of loss of memory is not Inco-operated in their day today’s modules (donna and Christopher,2015, P 163-177). Jacobs and Catherine in their chapter of how retrieval of information from memory enhances ability to learn is very important. In their work, the first stage of retrieval is the attempt get the answer from the bottom of the memory. Stage two begins when the answer is already available and ready to be presented. At this stage, appropriate connections become stronger while inappropriate connections are weakened. In conclusion they agree that, retrieval attempts indeed promote learning unlike retrieval success which does not. (2015, pg 283-294) In another experiment to determine the inference of in joint picture naming. Four experiments were done. Picture naming by one individual was found to be affected by the other individual as the belief was. An individual was found to take longer time to name a picture when they knew their partner was at the same time naming the same picture than when they thought that no partner was naming the same picture. However, picture naming was not affected by what a partner could say. These findings are similar to studies which showed that a speaker can be preparing his work so as to speak when another speaker is already speaking. (Gambia and Carveye 2015) Theory of associative learning heavily dwells on what happens when two different stimuli are presented together. In this animal study. The main focus lies on the advantages of concurrently influencing theories of associative learning. This study showed that attention is a major contributor in association formation. The differences came to whether it is the stimulus that contributes to the difference in attention. Recent researches have thus shown that proper attention is achieved when stimulus are paired up. (Bourton, 2001). Relation to the study at hand From the studies it showcases that learning is indeed a process, and it is affected by many other factors. Before learning is achieved , time factor is indeed a measure of what an individual achieves or is able to retrieve from there mind. Experiments and studies done also show initial stages of learning will show peak shifts with varying gradients. Increasing the time of both earning and retrieval from memory makes the learning process efficient. Conclusion 2 The peak-shift obtained were from the participants who were not able to acquire the rule-based approach performance during training. Therefore, for proper learning more time should be integrated into learning curriculums Those who did acquire the rule show the appropriate pattern throughout test. Remember, acquiring pattern is influenced by long training. References Aaron P. Blaisdell and Hanson, H. M. Effects of discrimination training on stimulus generalization. Journal of Experimental Psychology University of California, Los Angeles, United States1959, 58, 321-34. John M. Pearce. Theories of associative Learning in Animals. United Kingdom School of Psychology, Cardiff University. F.W. Jones and I.P.L. McLaren. Rules and Associations. University of Cambridge, Psychological Laboratory, Downing Street, Cambridge CB2 3EB. U.K. Shanks, D. R. (1995). The Psychology of Associative Learning. Cambridge, UK: Cambridge University Press. Livesey,E.J And McLaren. Journal Of Experimental Physiology: Animal Behavioral Process,Vol 35(4), Oct 2009,Pg 554-565. Forest CL, Mclaren, Aitken MR and Mackintosh NJ.Journal of Association and Preposition. pub 2013 Oct. Wills, S., & Mackintosh, N. J. (1998). Peak shift on an artificial dimension. Quarterly Journal of Experimental Psychology, 51B (1), 1-31. Todd R., Scahachtman and Steve S.Relly. Associative Learning and Conditional Theory: Human and Non-Human Applications.1999. Michael R.Rawson and David A, Mettler. Dictionary of Cognitive Science .University Of Cambridge.Pg 435. Shanks, D. R., & St John, M. F. (1994). Characteristics of dissociable human learning systems. Behavioral and Brain Sciences, 17, 367-447. Read More