I lied. The second article on vervets did not appear in Science, but in Current Biology. Other than that, I stand by my earlier statement: this is an unusual and thought provoking article, and it happens to be about vervet monkeys.
In this experiment, the authors created a locked container which contained appealing food. The subjects, a troop of vervet monkeys, could see inside and smell the food, but they could not get at it. Only a single low ranking member of the troop was trained in how to open the container.
Without needing to read the paper, I could tell you how things would start out. The dominant individuals in the group would mess around with the container, carrying out all kinds of violent acts in an attempt to force the thing open (which would all fail). Once the alpha got tired of this, it would go do something else and the second rank individual would make similar attempts to open the container. This is a common scene among the baboons, when they're trying to get into a locked storeroom or car.
In order for the monkeys to get at and eat the food inside the container, all dominant individuals would need to exhaust their own interest and wait. They were also required to stay a safe distance away from the container, somewhere between 10 and 15 meters. Only when the dominant individuals were a safe distance away would the monkey trained to open the container actually open the container.
It took most of these troops a few long trials to begin to understand that they would need to show restraint. But after they picked up this notion, the process went surprisingly smoothly and quickly, and the trained monkey would invariably be allowed to open the container. Unsurprisingly, the subject group that was experienced in raiding garbage bins and picnics took much longer to get over the fact that only a single low ranked individual could open the container. The authors go into great deal of game theory, but I will leave that by the wayside and get to the conclusions.
There are a number of fairly impressive accomplishments included in this paper. First, that the authors were able to train a wild vervet, to essentially be a confidant in their experiment. It is kind of incredible. Second, their main results show that these monkeys are capable of restraint on group and individual levels. Monkeys are notoriously bad at self-control, but this shows the power of reinforcement learning. I can't imagine any of the monkeys I've worked with, in captivity or the wild, being trained to show these levels of restraint. So kudos to the authors, and kudos to vervet monkeys.
In this experiment, the authors created a locked container which contained appealing food. The subjects, a troop of vervet monkeys, could see inside and smell the food, but they could not get at it. Only a single low ranking member of the troop was trained in how to open the container.
Without needing to read the paper, I could tell you how things would start out. The dominant individuals in the group would mess around with the container, carrying out all kinds of violent acts in an attempt to force the thing open (which would all fail). Once the alpha got tired of this, it would go do something else and the second rank individual would make similar attempts to open the container. This is a common scene among the baboons, when they're trying to get into a locked storeroom or car.
In order for the monkeys to get at and eat the food inside the container, all dominant individuals would need to exhaust their own interest and wait. They were also required to stay a safe distance away from the container, somewhere between 10 and 15 meters. Only when the dominant individuals were a safe distance away would the monkey trained to open the container actually open the container.
It took most of these troops a few long trials to begin to understand that they would need to show restraint. But after they picked up this notion, the process went surprisingly smoothly and quickly, and the trained monkey would invariably be allowed to open the container. Unsurprisingly, the subject group that was experienced in raiding garbage bins and picnics took much longer to get over the fact that only a single low ranked individual could open the container. The authors go into great deal of game theory, but I will leave that by the wayside and get to the conclusions.
There are a number of fairly impressive accomplishments included in this paper. First, that the authors were able to train a wild vervet, to essentially be a confidant in their experiment. It is kind of incredible. Second, their main results show that these monkeys are capable of restraint on group and individual levels. Monkeys are notoriously bad at self-control, but this shows the power of reinforcement learning. I can't imagine any of the monkeys I've worked with, in captivity or the wild, being trained to show these levels of restraint. So kudos to the authors, and kudos to vervet monkeys.
Posts
No comments:
Post a Comment