Introduction to Learning

Learning unlike most of the other concepts that are involved in this lab as well as the pre-cursing studies, can not be directly defined. Instead Learning has to be inferred from the change of behavior observed in the subject. A more or less permanent change implies the subject has "Learned" the desired response. Operant Conditioning is conditioning a specific subject to perform an Operant response to receive a reinforcer (such as food). An apparatus used to achieve this would be the "Skinner Box" or the "Davis universal feeder" (along with key light and response key) that was used in the Talton, Higa and Staddon study of 1999.

There are general concepts that are involved with most of the studies conducted on "Learning", such as Continuous Reinforcement. (CI) delivers a reinforcer whenever the subject performs the desired operant response, eventually creating high proportions of responses. Other forms of reinforcement that are involved within both this study as well as prior literature include; Fixed Interval (FI) trials in which a reinforcer is given on the first response after the set interval has elapsed. Variable Interval (VI) trials do not have a constant time interval, but instead vary around a designated mean. Finally the Extinction phase offers no reinforcement upon the achievement of the operant response, in an attempt to test the strength of the subjects previous training, or to reduce the occurrence of the behavior. Scalloping (or Inverse Scalloping) occurs during the FI training period. Early during this training Inverted scalloping occurs due to the subjects excitatory nature towards food, but as the subject begins to understand the length of the interval scalloping occurs. The subject responds most frequently only when there is a possibility of a reinforcer.

The study conducted by Talton et al (1999), used FI and VI schedules to explore the ability of goldfish to learn. They wanted to show that goldfish could achieve similar results as had been found during previous studies of rats and pigeons. Specifically to replicate studies that had shown sensitivity to temporal schedules in goldfish (Rozin, 1965; Wolf & Baer, 1963). The Talton et al (1999) study used 5 Shubunkin goldfish that were conditioned to perform a paddle pressing task. This task was conducted in the context of two different experiments.

In experiment one The goldfish were pre-conditioned for two weeks with FI schedules and were then studied under five different interval periods (all consisting of 25 trials). These were FI 60s, FI 240s, FI 30s, FI 60s, and finally a VI 60s. The observations from this study seemed to match the previous works. They showed the goldfish's seeming ability to learn temporal schedules for response and reinforcement. Specifically that the average wait times (latent periods) depended upon the length of the FI. The final FI 60s period and the VI 60s period did not differ greatly. These overall findings indicated that the fish learned to estimate the interval time of reinforcement and were therefore able to only increase their responses as they got closer to the interval time, and decrease them directly after receiving the reinforcer. Although these differences were not to the same scale as those visible in rats and pigeons they nonetheless provided evidence to support the temporal thought abilities of goldfish.

Their second experiment involved the use of a FI reinforcement procedure, in which they sporadically omitted the reinforcer. Although they omitted the reinforcer (food pellet) they continued to conduct the secondary stimuli during these omitted periods (ie. key light). These trials were conducted in FI 60s, they randomly selected 25% of the FI reinforcement periods and omitted the reinforcer (these omitted periods were designed so as not to have two omitted intervals in a row). The wait times for this experiment showed a slower increase then in experiment one. After the omitted periods the wait times were drastically decreased and total responses increased, (in comparison to those occurring after reinforced periods). This shows that these omission periods had an excitatory effect upon the goldfish. (Ie. They are expecting the reinforcer and therefore increase activity). These findings again support their hypothesis of goldfish's ability to learn temporal schedules for response and reinforcement.

The results from these conjoined experiments support the idea that, the conditioning patterns observed in both rats and pigeons can also be found in goldfish. These goldfish portray a similar form or ability for "Learn" different interval reinforcement schedules.

The Machado and Cevik study of 1998 used 6 pigeons to explore the processes of acquisition and extinction using periodic reinforcement. The same concept of having the subject achieve an operant response was used (pecking an illuminated key). To achieve this Machado and Cevik preconditioned them by using a continuous reinforcement schedule, and during the first phase of the experiment, exposed two groups of birds to two different fixed interval schedules (FI 40, FI 80). In the second phase they began with the FI schedules and then randomly began the extinction period.

During the first half of the FI training, this study found that there was little difference between the FI 40 and FI 80 groups. Their curves showed the same basic themes. Early after reinforcement the response rate decreased but as the interval continued and got closer to the reinforcement period there was a great increase in response rate. This data once graphed provides an exceptional example of scalloping. When coupled with the inverted scalloping that is shown in the early FI period and the straight line effect shown in the period of omittance it makes relevant comparative study. The speed of change between the steady and high response state is faster then the change from steady response to extinction. This can lead to the conclusion that different mental processes control the acquisition or extinction of a learned behavior.

A major difference between the Machado et al (1998) study and the current study is the form in which extinction is introduced. They introduce extinction mid-session during FI training, whereas the current study involves a separate phase of extinction.

The purpose of the present study is to explain the principles and techniques of Operant Conditioning, such as shaping the subjects responses by using Continuous Reinforcement. Then to introduce Intermittent Reinforcement through a Fixed Interval Schedule and finally to examine the process of Extinction

On the first day of the present study the rat will have been preconditioned with continuous reinforcement. It is likely that at first the subjects will take a bit to remember but will likely have a relatively high and constant