They showed 23 patients 80 photos of famous individuals and 80 photos of unknown individuals on a computer screen. They took care to ensure that the two samples matched each other in terms of age, gender, and ethnic origin, etc. With regard to the famous individuals, the psychologists conducted a pre-test on normal elderly individuals to ensure that the celebrities in question were very well-known. Alain Delon, for example. For the unknown individuals, a learning stage was required. The objective of the first step was to enable the patients to become familiar with the faces of the different individuals shown on the screen. “We showed the faces and then we asked the participants to indicate whether it was a man or a woman”, explains Sarah Genon. “We then began the stage of encoding the information in memory by giving the name of the person (Caroline Martin, for example) and in order to create an association between the face and the name in the patient’s memory, we asked questions such as "Do you think Caroline Martin is aptly named?”, and why”? Finally, we hid the name of the person on the computer screen where the photos appeared and asked the patient to recall  the name of the person immediately. In the case of a correct answer, we considered that the information was encoded and moved on to the next item; in the case of failure to provide a correct answer, we started the process again”.

A few minutes after the learning session, the participants in the experiment were again shown the photos of these fictional and famous individuals. On each occasion, four names were suggested to them. In order to avoid identifications that could be linked to a simple familiarity with regard to the phonology of the correct name, phonological distractors were introduced. For example, when the face of Alain Delon was presented, one of the four propositions was “Alain Seron”. In the same way, one of the suggestions for the “anonymous” Suzanne Firmin was Suzanne Quirtin. For each of the photos, the patients had to predict the possibility of a correct identification on a scale containing four levels: no chance (1), low likelihood (2), high likelihood (3) and certainty of success (4). “Level 1 referred to patients who considered that they definitely didn’t know the individual presented, as though they had never seen him or her”, says Sarah Genon. “In contrast, level 4 concerned patients who had direct access to the information in their memory and who were therefore capable of naming the face visible on the screen without any need whatsoever to give them a hint”. These two extreme categories, without any nuance, were not of any major interest to the researchers. They were interested in the two intermediate categories (low likelihood and high likelihood).

Failure

With regard to metacognition concerning episodic memory (fictional people), it emerged that, in relation to the group of elderly people unaffected by Alzheimer’s (a sample of 17 volunteers), the Alzheimer’s patients were more prone to pessimistic predictions (i.e. indicate low likelihood) of their ability to recognize the name, which they would in fact correctly identify a short time later – in a manner of speaking, they didn’t know that they knew. On the other hand, when metacognition concerned semantic memory (famous people), the patients knew that they had a good chance of recognizing the person. “They judged their semantic memory performance in a similar way to normal elderly individuals”, explains Sarah Genon.