Exploring the limits of consciousness
Michele Farisco is a researcher at the Centre for Research Ethics & Bioethics at Uppsala University.
What are the indicators of consciousness in humans and animals? And what about artificial intelligence – can computational systems develop consciousness? We asked Michele Farisco, a researcher in neuroethics at Uppsala university.
Farisco is a researcher at the Centre for Research Ethics & Bioethics at Uppsala University. He and his research group have developed a list of different indicators for consciousness. First of all, he points out that they use the term “indicators” rather than “criteria” since it is difficult to know for sure if another entity is conscious.
“We elaborated a list of indicators. One of these indicators is goal-directed behavior, which refers to when a subject acts in a specific way in order to reach a specific goal knowing that acting in that way is instrumental to achieve that goal. If a subject is able to behave in this way, it is likely that it is conscious.”
Another indicator on the list is brain anatomy and physiology. This indicator is based on the similarity between the architecture of our brain, which has been revealed as correlated to consciousness, and the architecture of the brain or homologous structures in other biological beings. If a similarity is found, it is likely that the being is conscious as well.
Episodic memory indicates consciousness
“One last example is episodic memory, which is the capacity of the agent to remember events experienced in a specific place and time. If you are able to attach yourself to a specific experience in the past, it is likely that you are conscious,” says Farisco.
Why is it important to pinpoint those indicators for consciousness?
“You can think about different scenarios where these indicators are particularly relevant. For instance, when patients are affected by disorders of consciousness, like coma or a vegetative state, which has been renamed as ‘unresponsive wakefulness syndrome’ and ‘minimally unconscious state patients’.
“In this case, you have subjects that are unable to show behaviorally their state of consciousness, so you need to think about indirect indicators, in order to get a more reliable diagnosis of these patients.”
Potential artificial consciousness
Another scenario, which is maybe more far-fetched, is about potential artificial consciousness. If you have an artificial system, very complex in terms of behavior or functions, you might wonder whether that system is conscious or not. To find out you need an indicator, something that can help you to attribute or to exclude consciousness.
Farisco and his research group recently published a paper in which they explored the connection between the evolution of the brain and the prospect of artificial consciousness. Photo: Getty Images
Farisco thinks that the rapid development of AI may have different and even opposite consequences on the way we think about consciousness.
“On one hand, we may still think that consciousness is something beyond the technical reproducibility. Let's say that consciousness is something specific to biological organisms, because you cannot reduce your subjective experience to computation and AI basically works on the basis of computation.”
In that case, it does not matter how much AI will evolve, because consciousness is on a completely different level or dimension – it lies beyond the possibility of AI.
“On the other hand, you may think that some aspects or dimensions of consciousness – specifically the more cognitive dimension – can completely be simulated or even emulated by artificial systems.”
Exploring the evolution of the brain
The capacity to consciously elaborate or exploit information in order to get specific goals, for instance, is something that you can replicate artificially. Maybe you can get even better results through a computational system, he states.
“Probably AI will never get the subjective experience of consciousness, but it might get another more cognitive dimension of consciousness. “
Farisco and his research group recently published a paper in which they explored the connection between the evolution of the brain and the prospect of artificial consciousness.
“Our starting point was how our brains have developed. And specifically, we highlighted some features of the brain that are crucial for our consciousness. We reflect upon whether it is possible that the present paradigm of AI will succeed in emulating consciousness.”
Their conclusion is that there are many shortcomings of the present approach that might be fixed in with neuromorphic AI. That is a specific line of research, which takes inspiration from the brain.
“We think this is a promising route to take, but still there is a long way to go. So in principle we do not exclude the possibility that we might have artificial consciousness, but we think that is a long way away from the present situation.”
Annica Hulth
The brain
Thanks to our brain, we can plan our actions, regulate emotions and solve complex problems. The brain also holds both positive and negative memories. But what happens when it malfunctions?