Pepperberg (1987)

Aim To see whether a parrot could use vocal labels to demonstrate a symbolic understanding of the concepts ‘same’ and ‘different’.

To see whether a parrot could use vocal labels to demonstrate a symbolic understanding of the concepts ‘same’ and ‘different’.

The concept of categorization persists in other species. In order for animals to identify items as same or different, they must recognise the category shared by items, and realise that this sameness can be applied to other, unfamiliar objects. This skill is known as 'symbolic representation'. There is evidence that birds have this skill as it helps in their survival. Doctor Irene Pepperberg is a research associate lecturing at Harvard University in Cambridge, MA. She was an animal psychologist.

Research Method and Variables
It was an animal case study involving one subject who was trained and tested over a couple of years.
IV = whether the object is familiar or new.
DV = whether the parrot responds correctly to the questions about ‘what’s same’ and ‘what’s different’ or not.

The subject was Alex, an African grey parrot. Sampled through the opportunity sampling technique.

He had been the focus of Pepperberg’s work since 1977. Alex had been involved in prior research on communication and cognition. He could name several colours, shapes, and materials. During the day (8 hours of experimental time), the parrot had free access to all areas of the laboratory, and at night he was confined to a white cage. When he had access to all areas of the lab, the request had to be contingent on the correct vocalisation (‘wanna go gym’). He could request fruits/vegetables/nuts. He was fed regularly and could request toys.

Alex was presented with two objects that could be differentiated based on 3 categories: colour, shape, and the material it was made from. E.g., Alex could be presented with a yellow wooden triangle and a blue wooden triangle. He would then be asked either ‘what’s same?’ or ‘what’s different?’ A correct response would only be recorded if he vocalised the appropriate category and not the 3 categories. So, in the above trial, the correct response would be ‘colour’ and not just ‘yellow/blue’.

Feature analysis involved in getting a correct response.
There were 4 processes Alex had to go through to get a correct response:

  1. Attend to multiple features of two different objects.
  2. From the vocal question, determine if the response is based on similarity or difference.
  3. Work out what is same or different.
  4. And vocally produce a category response.

Procedural Background
Alex had already been learning shapes, materials, and objects prior to the study. During the study, he enriched his learning. He was even tested on his novel objects that were not in his labelling repertoire meaning, and he would have to transfer his knowledge to other objects not used in training or presented to him before.
Training for Alex to acquire ‘colour and shape’ as labels took four months. To acquire ‘mah-mah’ it took nine months. The length of each session was dictated by Alex’s willingness to attend.

General training
The 'Model-Rival' technique used by Pepperberg is based on social learning. One human acts as a trainer and asks the second human questions about two presented objects. A reward is given for the right responses. The second human acts as a model to the parrot but, also as a rival for the trainer's attention. The roles of trainer/model well frequently reversed. When Alex gave the right vocalisation, he was given a reward/praise rather than to the model. Alex could see the label ‘colour’ and ‘shape’. Alex was trained on the third label “mah-mah,” which meant matter. Training sessions occurred 2 - 4 times per week from 5 minutes to 1 hour.

Trials were being carried out by secondary trainers who had never trained Alex. This was done so as to reduce any cueing effects from the original trainers. The ‘same/different’ questioning was incorporated into other tests sessions that were being conducted on him. On the day prior to the test, the principal trainer would list all possible objects that could be used for testing. A student who is not involved with Alex’s training would choose the question, form the ‘same/different’ object pairs, and randomly order the question. In a week, ‘same/different’ questions were asked between 1 & 4 times (so predictability was low).

Testing took place over 26 months. Pepperberg was present for all testing sessions. The same objects were never presented again during the test, so there was a single ‘first-trial’ response. If Alex answered the question incorrectly then the examiner removed the objects, emphatically said ‘NO!’ and turned their head away. But this was presented later till the correct response was achieved. This also taught Alex that an incorrect response meant that he did not get his desired object.

The principal trainer during the test procedure sat in the room with her back to Alex. She didn't look at Alex when objects were presented, she did not know what object was presented, and she repeated out loud what Alex said.

How test scores were calculated: total number of correct responses ÷ number of trials done.

Familiar objects:
• Alex correctly responded to 99 out of 129 trials (76.6%)
• He scored 69 out of 99 (69.7%) on first-trial responses.

Novel objects:
• Alex scored 96 out of 113 (85%) correct responses overall.
• Scored 79 out of 96 (82%) on first-trial responses.

• Alex scored 55/61 (90%) correct responses overall and 49/55 (89%) on first trial responses.


  • Parrots have the potential to demonstrate comprehension of the symbolic concepts ‘same’ and ‘different’.
  • They may also learn to respond to verbal questions to vocalise categorical labels.


  1. Measures were taken to ensure the study was valid by limiting researcher bias.
  2. The trainer conducting the trials had not trained Alex on the same/different task. This means that demand characteristics were limited.
  3. Quantitative data was collected which allows us to objectively analyse whether Alex could think abstractly.


  1. Alex was a lab parrot - this makes generalisation difficult because he is not representative of the general population of parrots.
  2. Alex suffered from boredom (which is why they switched up his tasks). The species is prone to self-injurious behaviour when bored (such as feather plucking).
  3. Ethics: being kept in a lab environment for an extended period of time can be considered unethical. Parrots are social species and as it was isolated in a cage at night-time, it could have caused Alex stress.

Application to Everyday Life
Learning techniques have enabled us to see that other animals have the skill of symbolic representation and can use these concepts in novel situations.

Nature vs. Nurture
Operant conditioning and social learning demonstrated are nurture-based approaches to learning. Although, the differences in parrots and humans abilities are due to inborn traits.

Individual and Situational Explanation to Behaviour
Alex’s abilities in the familiar and noble testing showed differences (higher success on unfamiliar object pairs). This suggests that situational factors (familiarity of objects) affect an animals ability to answer right.

Use of Animals in Research
Advantage: cheap and easy to care for; responds well in vocabulary exercises; no evidence of Alex being harmed as a result of the learning techniques; easier to conduct a longitudinal study with parrots than with humans. Alex was motivated by receiving rewards — motivating humans would be harder.

Disadvantage: Parrots are unable to explain in-depth reasonings. There are concerns about keeping an animal in captivity and engaging it in artificial tasks.