Wednesday, December 5, 2012

Reward Probability and the Variability of Foraging Behavior in Rats

    This week's blog post is about a topic and paper from my very own professor, Dr. Aaron Blaisdell and his colleague at the University of California, Los Angeles, Dr. Stahlman.  The primary research study is called "Reward Probability and the Variability of Foraging Behavior in Rats" and can be found in the International Journal of Comparative Psychology from 2011. 
     The main purpose of this study was to build upon the idea of exploitation vs. exploration for food foraging behavior in rats.  Research has previously been conducted in this area involving nose-poking, bar-pressing and bar-sequence variability in rats.  When a source is continuously providing reinforcement and reward, then we will continue to go to that source and exploit it until something changes.  The change may come about with a reduced probability of reward as resources dwindle in nature, leading to more variability in the form of responding.  This is the typical procedure that occurs when a behavior is being extinguished in a lab.  Reward will no longer be as available as was previously learned (the probability of reward is reduced), until the participants change their behavior and a new response could be conditioned.  In nature, this is seen as well, as food or resources can be exploited and used up, until very little remains at the source.  This would require animals to begin searching for new areas of food, and change their typical behavior so that they can explore their surroundings.  The behavior of variability and exploration increases the probability that an organism is going to discover and exploit more highly rewarded behaviors.
      In this experiment, rats were investigated for their behavioral variability when the probability of food reward was alternated between 100% and 20% in an open field.  10 female rats were used in this experiment and were placed in plastic tubes in a vivarium (an enclosure meant to keep animals).  The open field apparatus was made up of a flat piece of wood placed 94cm above the floor with sixteen cups in a 4x4 arrangement arranged so that their top extended only 2cm above the surface of the open field.  The cups were then filled with wood shavings and possibly a positive reinforcement - Cocoa Puffs - above or below the wood shavings.
      The procedure consisted of habituation training sessions, followed by training and testing sessions.  For two habituation sessions, cereal was scattered across the open field and rats were allowed to explore for ten minutes.  I presume that this was done so that rats become familiar with the idea of searching for food and the apparatus, and so that they would not be surprised about the sources of food during any training or testing sessions.  This was followed by eight training sessions, each session consisting of two HI trials and two LO trials.  The design consisted of wooden landmarks(LM) with a separate LM for the HI and LO trials.  The landmarks were meant to point rats in the direction of food and signal the probability of food.  The trials ended when all the food had been consumed or after 3 minutes.  The testing session followed this and consisted of LO trials with only 20% probability of finding food and HI trials that remained at 100%.  Trials were terminated when rats searched the goal cup or after 3 minutes.  The interresponse distance was measured, which served as a measure of spatial variability in responding.  In addition, the number of mean searches was measured.
         The results showed that there was no significant effect of mean searchers on HI vs LO trials, but there was a trend toward more searchers on LO trials (2.8 vs 2.1).  Results also showed that over the entire testing session, the mean number of searches decreased over time, leading to decreased variability for both the HI and LO trials.  The distribution of search frequencies was wider with LO than HI trials, showing that behavioral variability was greater with a reduced expectation of reward.  Finally, the last important finding from the study was that out of 184 trials, rats failed to find the cup with the food in it (goal cup) only 1% of the time in the HI trials and 7% of the time in the LO trials.
         The significance and application of this research cannot be overstated.  This research is consistent with other findings and also expands upon previous research with lever-pressing in rats or screen-pecking in pigeons with a more environmentally applicable demonstration.  Animals in the wild will utilize the techniques that were demonstrated in this experiment because it is essential to their survival  These findings support the behavior systems view of foraging behavior showing that animals may engage in less consummatory mode behaviors and more in appetitive (search) mode behaviors when the expectation of reward is low.  If one source isn't leading to results, then animals will conduct trial-and-error and vary their responses until something works.  When the expectation of reward is high, then animals will continue to exploit a source until it is depleted.  This is a naturally adaptive trait, and now it has been proven to occur in an open field study with food reinforcement.
          As the researchers allude to, further research conducted in this field would include discovering the role of variability and how it leads to actions that are highly-rewarded and what the process is in that.  In addition, I would like to see experimenters look at how much effort would animals be willing to input in exploiting resources.  For example, let's say animals have a steady source of food that provides a 70% reward, but there is a nearby food source that provides 90%, however, it would require extra effort/energy to get to or to unlock and begin exploiting.  Are animals always comfortable in the source that they are currently exploiting until it is depleted, or are they always looking for the highest probability reward system?






                                                                            References

Stahlman, W. David, and Aaron P. Blaisdell.  "Reward Probability and the Variability of Foraging   
           Behavior in Rats."  International Journal of Comparative Psychology 24 (2011): 168-176.

No comments:

Post a Comment