New research has brought scientists another step closer to creating an artificial neural network with metamemory.
Metamemory is the process by which a person has cognitive awareness of their own memory capabilities and uses this "metacognition" to adjust their behaviour.
Using a computer-based evolution experiment, a team from the Graduate School of Informatics at Nagoya University in Japan's Chikusa-ku, made use of the recent breakthroughs in designing artificial intelligence (AI) technology using neural networks that imitate human brain circuits.
The findings of the study, 'Creating artificial intelligence that behaves more like a human by ’knowing that it-=knows’,' have been published in the online edition of the international scientific journal Scientific Reports.
The Nagoya team comprising Professor Takaya Arita, Yusuke Yamato, and Reiji Suzuki created an artificial neural network model that was able to carry out a delayed ‘matching-to-sample’ task and subsequently analysed its behaviour.
“To clarify the evolutionary basis of the human mind and consciousness, it is important to understand metamemory. A truly human-like artificial intelligence, which can be interacted with and enjoyed like a family member in a person’s home, is an artificial intelligence that has a certain amount of metamemory, as it has the ability to remember things that it once heard or learned,” explained lead author Professor Arita.
The ‘delayed matching-to-sample task’ is a typical go-to scheme when researchers study metamemory. In the case of humans, this task presupposes seeing an object, remembering it, and then participating in a test to select the aforementioned object from a multiple selection of things.
Throughout the test, for which a human would naturally use their metamemory, correct answers are rewarded and wrong answers punished. Previous studies revealed that monkeys could also perform this task.
The model employed by the research team from Nagoya was able to evolve to the point that it performed the tasks with success similar to that of monkeys. The neural network created by the study could examine its memories, retain them and even separate outputs without any help from the researchers.
“The need for metamemory depends on the user's environment. Therefore, it is important for artificial intelligence to have a metamemory that adapts to its environment by learning and evolving. The key point is that the artificial intelligence learns and evolves to create a metamemory that adapts to its environment,” said Professor Arita.
This achievement, according to the team, may be instrumental in providing clues to the eventual “realisation of artificial intelligence with a ‘human-like mind’ and even consciousness.”