The gaze and babbling of babies are part of the material with which a team from the National Center for Artificial Intelligence, Cenia, is working to create the world’s first early cognitive development curve. Comparable to the one used to measure aspects such as height and weight of children. The research is led by Marcela Peña, alternate director and scientist at Cenia, a pediatrician, with a PhD in Cognitive Sciences and Psycholinguistics from the Ecole des Hautes Etudes Sociales in Paris. She is providing data and materials she has collected during her career to Cenia’s engineers and mathematicians so that through Artificial Intelligence resources they can obtain the expected indicators. Basically, they are applying machine learning techniques.
This consists of classifying the materials available in categories that allow the construction of developmental curves. These should make it possible to identify the expected data and compare them with the data observed in infants and preschoolers who are in a stage of accelerated cognitive development which refers to the growth of a child’s ability to think and reason. It occurs in different ways from birth to adolescence, although its exponential nature is recognized in the first 3 years. The objective is to reach conclusions that allow the construction of development indicators with a high level of predictability, such as those that exist in relation to factors such as weight and height.
In collaboration with France, Canada and Spain
The idea is that the early cognitive development curves will help to detect difficulties in a timely manner and generate interventions to overcome them. Dr. Peña explains the scope of the model they are configuring: “When you take your young child to the doctor, they tell you, if he/she is 4 months old he/she should weigh between a certain amount and another, at 6 months old he/she should weigh between other amounts because the child grows. If your child is not in those ranges, either below or above, you have to do research and/or interventions to fix the eventual problems that come from being outside the expected weight ranges.”
She emphasizes that such a curve does not currently exist in the world in relation to early cognitive development: “in language, we know more or less what children should speak at a certain age, but we do not know how they should speak or babble. We have no precise scales for almost anything in relation to cognitive development.”
In this regard, she explains: “For example, if all children babble in a similar way, in terms of type of vocalization, melody, duration, intensity, etc., if we can build a curve with these data we can detect if any infant babbles atypically, which will alert us about the need to research and intervene early and in a timely manner. She says that some observations of these characteristics have been made in relation to the crying of infants at birth and it is known that if they cry in a different way from the standard, it can mean a disorder that should be treated.
In relation to the material available to them, Dr. Peña mentions “voice records of children who have participated in our vocalization studies, we also have many records of children performing linguistic tasks, and visual records in which the gaze is detected. For example, if a child prefers to look at faces that have an evident emotional expression, versus those that do not.”
Recently, they have worked with children around three years old, identifying some characteristics of their behavior: “When they come to the first session they don’t talk, because they don’t know the evaluator. Maybe because they say to themselves ‘You know what? I don’t want to talk.’ After the second session they say to themselves ‘This person is not so strange, I’m going to tell him something.’ From about the fourth session onwards they go off and talk, construct new and complex sentences and play with language.”
From the set of data obtained, they choose certain aspects towards which they will direct the search through Artificial Intelligence: “For example, one could determine whether what matters in the voice is the intonation, or the number of pauses, or the content, or the repetition of words.” From this point of view, the research being carried out is also pioneering, since it involves modeling language through speech, in circumstances where practically all the research that has been done through artificial intelligence up until now has been based on texts. Towards the future, Dr. Peña poses a greater challenge, to model the internal language that human beings incessantly develop in thought. She believes that the only method that could achieve results in this field is Artificial Intelligence.
Recently, they have worked with children around three years old, identifying some characteristics of their behavior: “When they come to the first session they don’t talk, because they don’t know the evaluator. Maybe because they say to themselves ‘You know what? I don’t want to talk.’ After the second session they say to themselves ‘This person is not so strange, I’m going to tell him something.’ From about the fourth session onwards they go off and talk, construct new and complex sentences and play with language.”
From the set of data obtained, they choose certain aspects towards which they will direct the search through Artificial Intelligence: “For example, one could determine whether what matters in the voice is the intonation, or the number of pauses, or the content, or the repetition of words.” From this point of view, the research being carried out is also pioneering, since it involves modeling language through speech, in circumstances where practically all the research that has been done through artificial intelligence up until now has been based on texts. Towards the future, Dr. Peña poses a greater challenge, to model the internal language that human beings incessantly develop in thought. She believes that the only method that could achieve results in this field is Artificial Intelligence.
Not with monkeys
The researcher recognizes that her vocational interests go back to her childhood and always strongly influenced by her emotions. It literally all started by playing, with a mother who was very playful. “We played that I was the doctor, and my brothers, one an engineer and the other a lawyer. Then we studied and although she never asked us for anything now I am a doctor, one of my brothers is an engineer and the other a lawyer. So somehow we thought we were still playing.”
When she left high school she did not have a definite choice for university studies. Those were the times of the Academic Aptitude Test and there were 17 careers that she could choose from. She chose all of them. She chose medicine and was accepted, but she believes she would have liked to “do anything I was given the opportunity to do. My parents were not rich, they were hard-working, super enthusiastic, loving, respectful and very cultured. They always took care of us and we were able to do what we wanted.”
Not with monkeys
The researcher recognizes that her vocational interests go back to her childhood and always strongly influenced by her emotions. It literally all started by playing, with a mother who was very playful. “We played that I was the doctor, and my brothers, one an engineer and the other a lawyer. Then we studied and although she never asked us for anything now I am a doctor, one of my brothers is an engineer and the other a lawyer. So somehow we thought we were still playing.”
When she left high school she did not have a definite choice for university studies. Those were the times of the Academic Aptitude Test and there were 17 careers that she could choose from. She chose all of them. She chose medicine and was accepted, but she believes she would have liked to “do anything I was given the opportunity to do. My parents were not rich, they were hard-working, super enthusiastic, loving, respectful and very cultured. They always took care of us and we were able to do what we wanted.”
With this concern, she started to find out where she could find more information on the subject. She obtained the President of the Republic scholarship and began to look for where to go. In this search she went through Yale and MIT. But everywhere she went she was told that to fulfill her purpose she had to work with monkeys. “There was no major research being done with premature children,” she says, “and working with monkeys was not possible for me. It was like looking at little children. When you came in, they would cling to the fence and look you in the eye. I felt like opening the door to let them escape, so it didn’t work out.”
20 million words less
After this experience she came into contact with one of the world’s leading authorities on the study of early cognitive development, Professor Jacques Mehler, who accepted her for further studies in France: “He told me wonderful things about children who knew how to think, even if they couldn’t speak. He passed me a book called ‘Born Knowing.’ I read it and was fascinated. I realized that for the first time I liked something desperately.”
She received her doctorate from the Ecole des Hautes Etudes Sociales in Paris, did a post-doctorate in Italy and returned to Chile. She built a team very committed to children, composed, among others, by Dr. Enrica Pittaluga, the speech therapist Orieta Palacios, the psychopedagogue Consuelo de la Riva, a large group of early childhood educators, psychologists and a dozen students.
She has worked with children who are treated at the Hospital Dr. Sótero del Río and who present different degrees of vulnerability. There he began a trajectory that is linked to the purpose of his current incursion into the field of Artificial Intelligence.
She was impressed by the fact that some of the children she worked with had up to 20 million fewer words than those from more affluent sectors: “I said to myself, how is this possible? How can someone in their environment not talk to them to increase their vocabulary? It was said that they could acquire it through television. But it has been proven that this media does not fulfill this function, fundamentally because TV is not interactive. It is not comparable to the effect of a person talking to the child.”
Her reflection led her to where she is now: “We have to do something to help develop language, so that girls and boys increase their vocabulary and communication skills. We have to make it so that people don’t have vocabulary limitations. Because they do have the capabilities to develop a large vocabulary. To that end, we must start early. Children don’t talk much until they are about two years old and suddenly they discover this tool, use it for everything and play with it. We have to explore how we can take advantage of that great ability to speak and understand to format their thinking, so that they can explain in words their emotions, their ideas, their beliefs, their judgments, their decisions, and they can fulfill themselves through language.”
By Gonzalo Rojas Donoso. Llambías Comunicaciones.
