Emotion and robotics, focus on the research work of Professor Lola Cañamero

Can you tell us more about yourself and your career?

I first studied philosophy in Spain for seven years. I was particularly interested in how we get to know and interact with the world. How, as entities, do we decide and do things by interacting with the world?
It was during my studies in philosophy that I encountered artificial intelligence. I thought it was a good way to approach the philosophical questions and try to see how the answers worked. In fact, with artificial intelligence (AI) we can program hypotheses and carry out tests. We don't just stop at conceptual thinking. This is how I started doing AI and started a PhD on this subject. To do this, I came to France to the Computer Science Research Laboratory (LRI) of the University of Paris Sud where I worked on issues of learning and modeling knowledge to put them into the machine so that it can reason. I did a thesis on the recognition of action plans of other agents based on their observed behavior, applied to the field of car driving, in the perspective of what is called classical AI, i.e. artificial intelligence which consists in modeling reasoning processes as we perceive them when we think. 

At the same time, I spent a few months in the AI laboratory of the Vrije Universiteit Brussel, led by Luc Steels, where I met Rodney Brooks, who was on sabbatical at the time. Brooks, with Luc Steels, founded another, very different approach to AI: AI embodied. With this approach, weare interested in modeling how intelligence is the result of interaction with the real world and the body. Movements, perception, action and their connection have an impact on the cognitive process. Intelligence is a phenomenon that emerges from interactions and that fascinated me. I then went to do a postdoc in the United States with Brooks at MIT in 1995-1996, and that's when I started working on emotions. I thus discovered innovative research in neuroscience that dealt with the role that emotions play in cognitive processes such as decision-making or social interactions. Initially, I was interested in this in the context of decision making for autonomous robots. It was in 1995 and since then I have continued to work on this subject.

I then spent most of my career in England at the University of Hertfordshire. Between 2001 and 2020, I led a team of researchers to work on the modeling of emotions and its interaction with cognition. I had the opportunity to work with a lot of researchers, doctoral students, and post doctoral students who have made progress with me on this subject.
This year, I joined CY Cergy Paris University for a senior scientific excellence chair in robotics and neurosciences in order to continue my research and to approach new facets of it.

So what are the themes and issues of your research work under the Chair?

I am currently working on the modeling of affective cognition. In fact, it involves looking at the interactions of affective and cognitive phenomena and I'm modeling that in robots.
As part of the chair, I focus on the interactions between affective phenomena such as emotions or motivation and cognitive processes such as decision-making or cognition and social interaction, drawing inspiration from neurosciences and other disciplines such as ethology and cybernetics, and of course philosophy, social sciences and humanities.
Emotions are very complex and multi-faceted phenomena. There is of course the expression of emotions in the context of communication with others, through facial behavior, movements, the body, as well as the perception of emotions by others, humans or even robots, that recognize the emotions of humans to better interact with them.

Additionally, there is a less covered theme which is the interaction of emotions with cognitive phenomena so that the robot can make autonomous decisions and be adapted to what is happening around it based on the humans with whom it has to interact. For example, in the physical world if there is a danger, what elements should be taken into account and also how to interact with humans in these situations?
I would also like to go deeper into more social elements. Indeed, during decision-making, cultural constraints, elements linked to social relationships such as hierarchy, are important. In order to do this, I have studied the development of emotions, for example by looking at attachment ties and using the attachment and interactions between a baby and his parents as a model. The goal is to be able to model these interactions so that a robot can learn directly from the humans with whom it  has to live, interact or work as a child learns and interacts with its parents. I thus sought to understand what happens in the brain to create these bonds of attachment and how this impacts the learning and the memorization of things.

However, at this stage I have not taken into account the cultural differences. There are still very different cultural norms, especially in terms of appropriate and inappropriate behaviors in different situations. I would like to develop this part on cultural links while remaining closely linked to neuroscience. I would like to bring the humanities and neurosciences a little closer together through robotics.

Concretely, what are the points that you are going to study and develop in the coming months?

From the scientific point of view, I must address the different aspects of the phenomenon of emotions in parallel. I will thus try to understand an emotion and how it is relevant for us and for a robot by studying the different underlying mechanisms that trigger an emotion and its equivalent in a robot.
I will also need to understand the value of these behaviors. Are they adaptive, positive, or harmful to agents in solving a problem? This is called the adaptive value of emotions. Does what makes sense for humans also make sense for robots? Or should the functions be changed?

I will also be working on emotional and cognitive development, to see why emotions are different depending on the attachment bond built, its context and interactions with others, and whether the emotions are good or bad. Finally, the emotions we have are the result of an evolution of the species and there are elements that can have a positive or negative function depending on the context. For example, in the expression of an emotion there are elements that are useless and have remained, but may be irrelevant and therefore irrelevant for a robot.
These are many different perspectives to try to figure out what emotions are and what should be modeled in robots that interact with us.

Can you tell us more about robotics? What do you mean by robot?

Robotics today takes many forms. When we talk about robots, people often think of science fiction, of robots who want to dominate the world. But this is not my vision. Today we have smaller robots out there that can perform very specific services and we sometimes have them at home. These are not necessarily machines that have a humanoid form. They are in our lives and will continue to be more and more.

How do you fit your research into issues of society or the business world?

So far I have mainly done fundamental research to understand the principles and set up models. But I have also been in contact with institutions to develop robots in the field of services or health and to elaborate models. For example, I worked with psychiatrists on mental disorders to see how the robotic model could help to understand these disorders. Another example, I have developed companion robots for diabetic children as part of a European project. My research and its modeling can be applied to issues in society or the business world and it is important today that companies rely on solid research done by researchers to respond to the challenges of tomorrow.