Avian Learning Experiment

from Wikipedia, the free encyclopedia

The Avian Learning Experiment (English for "bird learning experiment ") is a long-term project carried out by the university animal psychologist Irene Pepperberg , in which the learning and thinking abilities of African gray parrots were explored. The focus is on cognitive skills in naming and grammar when parrots use human language. The best-known African gray parrot in her group was Alex .

The aim of the work

The fundamental question of the investigations was whether an animal like a parrot is actually able to communicate consciously, or whether it is only conditioned to external stimuli and reacts with a behavior that promises the greatest reward.

The tests carried out should clarify whether, with appropriate training, a parrot can on the one hand learn to classify objects into categories and, on the other hand, understand the principle of like and unlike objects. These results could be used to infer one's own language behavior, or at least the system for it.

Classification of categories

Categorical class formation
In categorical class formation , things are classified in an associated category based on their properties . A blue wooden cube would have its own category that is valid for all the same blue wooden cubes.
Stimulus generalization
In the case of stimulus generalization, for example, all cubes, regardless of their color, would be classified in the category of cubes.

In the following, it should be tested whether Alex has the ability to form individual categories or whether he classifies both into the same category on the basis of a given stimulus that distinguishes two similar objects.

Experimental setup

The so-called model / rival technique (German: example / rival) was used. There is an instructor who asks a second instructor - the rival - to do something. This rival will be rewarded or scolded depending on his reaction. The parrot observes this and recognizes the system according to which there are rewards.

In the specific case with Alex, this meant that Alex should differentiate between objects based on categories, namely according to shape, color and texture. In contrast, there was the classification according to the stimulus generalization described above, according to which all objects that have a certain characteristic automatically fall into the same category.

Results

Since Alex was able to divide the same object into different categories depending on the question asked (e.g. with regard to the color on the one hand and the shape on the other hand) and to express this through a sound feedback - through "vocal labels", it was generally possible for him to differentiate between categories.

The mistakes Alex made were as follows:

  1. It was considered a mistake if Alex chose the right category but named the wrong color.
  2. Alex's mistakes were seldom completely wrong, his mistakes can be divided into three types:
    1. Was it about mixed vocabulary between colors and shapes (for example "roween", a word made up of rawhide and green). These mistakes made up about 4% of his mistakes.
    2. Sometimes Alex skipped a marker. This error accounted for about 8%.
    3. The most common error occurred when Alex recognized one marker correctly and not the other, for example choosing the right category but choosing the wrong label (example: "green" instead of "blue"), or when he read incomplete labels ("wooden corner" instead of “wooden triangle”). These errors together made up about 12%.

An extension of the experiment

Pepperberg wanted to find out whether it was possible for Alex to transfer what he had learned from one experiment to a new one. In tests in which Alex had to distinguish between five different colors, five different shapes and three different materials, he was 80% correct.

In tests that asked about known colors and shapes, two new objects and two known objects, his answers to the question about color were 84.7% correct and when asked about shape 83.7% were correct.

Since the reward in the tests was that he was allowed to take whatever item the question was about in his beak and then drop it - which apparently pleased him - one might suspect that it was newly introduced Objects caught Alex's interest more than familiar ones.

Results

Alex was able to understand a whole range of verbally expressed questions and also recognize the relevant category of the question and the object, which could be classified with several possibilities. He was also able to correctly categorize this instance .

Yet his skills were still very rudimentary when compared to non-human primates or marine mammals .

Alex was not able to describe the difference between two objects on his own, nor was he able to pick out a particular symbol from a crowd. Alex was also able to answer the question “What color is the object?”, But he did not understand the concept of independent differentiation and the subsequent description.

Understanding the principle of equal / unequal

Pepperberg investigated whether Alex was able to understand the concept of equal and unequal. The state of research at the time of the experiments was that it was assumed that animals could learn to distinguish between two objects through training, but that they did not have to understand the concept of equal and unequal .

Experimental setup

The M / R technique was slightly modified. One difficulty for Alex was that he now had to "rethink" the training methods he had already learned.

In the experiments, Alex had to answer the questions "What same?" And "What different?" And name the relevant properties of the, now two, objects. The categories that Alex still had available in his previous knowledge were:

  1. five colors
  2. some forms
  3. four materials
  4. some metallic objects (key, chain etc.)
  5. Understanding of abstract categories ("What color?" "What shape?")
  6. a variety of other terms (food, places, amounts, etc.)

He had to label them with the correct “vocal label” in the relevant category.

Results

After nine months, Alex was 76% able to verbally express equality or inequality on known objects. In the case of unknown objects, even 85%, which in turn indicated that he must have understood the concept, since he delivered better results with "untrained objects". The test questions as to whether Alex had answered the correct question were 90% proof of the correctness of his answer. Pepperberg attributed the comparatively long learning time of nine months to the fact that Alex sometimes had difficulty uttering some sounds and, in addition to these attempts, he also learned to count. He was also able to answer the question “How many?” Later.

It is interesting that, compared to experiments with chimpanzees , Alex was able to name the exact characteristics that made the difference, and not just to indicate in general whether the objects were the same or not. It must be said, however, that Irene Pepperberg assumed that the principle of differentiating between equal and unequal would have to be present in parrots up to a certain degree from birth , since they are in the wild e.g. B. must distinguish between chants and food. Thus, the tested object differentiation would be a vital property that would be present in parrots in particular, which in turn would make a comparison with some other species more difficult, which have to resort to it less or not at all.

Language ability

After the 19 years of training Alex had its own vocabulary of 200 words he could utter, and a vocabulary of about 500 words that he could understand. In addition, he could - to a limited extent - count and express his own wishes. For example, if Alex no longer liked the question-and-answer game, he would say, “I'm gonna go away”, and if the researcher was angry, Alex would try the situation with “I'm sorry” (I am sorry sorry) to defuse. If he had asked for a banana via “Wanna banana” but was offered a nut, he would look in front of himself in silence, repeat the request for a banana or take the nut and throw it at the researcher with his beak. Alex's understanding of numbers was in the range of under 7. When asked how many objects of a certain color were on a board, he gave the correct answer 80% of the time.

state of research

Preliminary research suggests that Alex was able to translate the concept of four objects (e.g. four blue wool balls on a board) onto four piano notes. Pepperberg tried to teach him the concept of the written number "4" as a representation of the number "four".

In July 2005, Pepperberg reported that Alex understood the idea of ​​the number zero . Apparently he could also add .

Pepperberg tried to teach him phonemes of the English language with the intention of being able to relate written and spoken words. According to her, however, Alex did not use human language, but a complex communication scheme.

literature

  • Pepperberg, IM: Alex and me. The unique friendship between a Harvard researcher and the smartest bird in the world . mvg Verlag, 2009, ISBN 3-86882-026-4

Web links

Individual evidence

  1. ↑ Pepperberg 2009, p. 70 (English edition)
  2. ↑ Ask the Scientists: Irene Pepperberg Q&A . Retrieved September 11, 2007.
  3. ↑ Researchers explore whether parrot has concept of zero . Archived from the original on September 3, 2007. Retrieved September 11, 2007.
  4. ↑ Michael Miersch: In the beginning there was the word. In: welt.de . October 5, 2009, accessed October 7, 2018 .