This article was originally published by the ABC in 2015 and was written by Olivia Willis.

It sounds like an episode of Dr Who, but the possibility of emulating mental content has prompted a previously unthinkable question: could you really upload your mind?

'Today there are people who really think this idea is plausible, and that by means of technological developments, we will soon be capable to transfer our minds to a computer and therefore achieve some kind of immortality,' says Max Cappuccio, a philosopher of mind at United Arab Emirates University.

When I want to upload my mind to a computer, I’m not just interested in having a replica or a duplicate of my mind, but I want it to be exactly my mind that is transferring to the computer.

Max Cappuccio, Philosopher

'This requires that our mind is compatible with a computational device ... or at least offers some of the basic features that would allow it to be emulated by a computer.'

Setting aside the enormous technological issues, Cappuccio is struck by the philosophical questions—or rather, challenges—uploading the mind might pose.

'When I want to upload my mind to a computer, I'm not just interested in having a replica or a duplicate of my mind, but I want it to be exactly my mind that is transferring to the computer.

'Because of course that's my main interest, I want to become immortal.'

Cappuccio says that in order to achieve that goal, the hypothetical mind algorithm can't simply be qualitatively identical to the original, it must also be numerically identical—it must be 'one and the same' with the original.

'Now, there's a little problem with that,' says Cappuccio.

'In order to keep the numerical identity, we must deal with something that is individuated.'

The principle of individuation, as imagined by Aristotle, is a criterion which supposedly determines whether we can have more than one of any one thing, based its material qualities and contingencies.

According to the theory, nothing can have an individual identity—and therefore become numerically identical—unless it holds material qualities and properties.

'Let's say I'm thinking of Pythagoras' theorem, and then you are also thinking of Pythagoras' theorem. Is it one and the same, or just two tokens of the same type?' asks Cuppuccio.

'It's pretty much impossible to answer this question, because there's no way simply to tell whether abstract immaterial entities are one of the same, or two different things.'

According to Cappuccio, this same argument applies to algorithms, including the 'mind upload' process.

'These [mental] operations are not real objects, they're relationships between objects, qualities or properties. But they're not real things in themselves,' says Cappuccio.

Cappuccio says it's a case of wanting to have our cake and eat it too.

'On the one hand, we are saying that minds are immaterial, and this is a requirement of mind upload because if minds were anything material, then basically we couldn't leave our body and gain what trans-humanists call this kind of liberation from the body.

'On the other hand, if we want to keep numerical identity, we must have something that preserves this principle of individuation. That is absolutely necessary, at least from the point of Aristotle, who says that nothing can be exactly one and the same unless it is something that is material.'

Here, Cappuccio highlights a blatant contradiction: we want our mind to be something that is free from material character, but at the same time we want it to be something that is individually identical to itself.

As Cappuccio sees it, the idea of the 'mind upload' is classical Cartesianism, which says the mind is a non-physical substance wholly separate from the corporeal body.

'There is a sort of Cartesian dualism, in the sense that we are postulating that the mind is not just a function,  but a substance that exists in itself and by itself independently of the body.'

This conception of mind runs into difficulties in the 21st century.

'If we want to provide a purely scientific account of the mind—our approach to be enlightened by the basic principles of cognitive science—then we have to endorse a naturalistic point of view,' points out Cappuccio. 'We cannot mix natural and supernatural entities, right?'

So is there a theory that can deal with the naturalism modern science points to?

Much work in recent years has been undertaken on the extended, embedded mind—the idea that what you think takes place within the confines of your skull is actually distributed more widely.

Cappuccio sees one particular branch of this new physical account as offering a better route to how things really are: radically enactive cognition (REC) theory. And it's implacably opposed to mind upload. 

'I believe they are archenemies. They are two opposite points of view, and they exclude one another,'  says Cappuccio.

'Mind upload essentially requires that our mind is comprised of content for representation. REC, on the other hand, asserts the opposite. REC says basic minds cannot have and don't need representational contents.'

Instead, Cappuccio says representational content belongs in socio-cultural and linguistic domains, because it is the product of human social practice and communication capability.

'It absolutely makes sense to talk about representations when we talk about a work of art, a story being narrated, a book, a novel and so on. That's a representation of our life. That's a specific place in which representations exist.'

But according to the REC theory, we cannot attribute these representational functions to the basic functioning of our brain like mind upload does.

The battle between the two theories is not just a thought experiment for idle moments. It has some major ramifications in one very important modern arena—the drive to create artificial intelligence.

'The root of intelligence is in … this immediate responsiveness and capability to adapt. And this adaptability cannot exist if you're not embodied if you're not situated in real life contingencies,' says Cappuccio.

Luc Steels, one of the world's leading robotics researchers, is putting the theory to the test.  

He's trying to get robots to talk to one another, and in the process is putting into practice the re-think Cappuccio entertains: the idea that the mind is perhaps not where we think it is.

He says it all began with some early work with two electronic tortoises called Elmer and Elsie, who were pitted against each other for scarce resources.

'One experiment that totally surprised me was that we started these robots with the same project. They had some way to know when to go to the charging station and whether they could do more work to learn about their environment to navigate around.'

In the beginning, the robots were doing everything according to their programming.

'And then at some point, we noticed that some of these robots became kind of masters of the other robots, in the sense that they worked much, much less … than these other robots, which were sort of enslaved to do more work.'

According to Steels, this division of labour and apparent social inequality was not part of the programming.

'Nevertheless, this kind of phenomenon emerged in this population,' he says.

The program's learning mechanism was 'very simple', according to Steels, and the results indicate critical lessons about the significance of environmental factors.

'There was just one parameter … inside the program, which the robot itself can change, and so it is increasing or decreasing that parameter based on its experience with the environment.

'This, I think, is a major, important point,' says Steels.

'The complexity we see is not due to complexity in the brains of these robots, but is actually due to this interaction between simple behaviours, interaction with the environment and interaction with other robots.'

According to Steels, some earlier robotics studies were also motivated by a desire to understand the origins of complex behaviour.

But unlike previous experiments that focused on complexities arising from within the agents themselves, Steels was more interested in exploring how environmental factors shaped the robots behaviour.

'This is how I then came back to my interest in language and trying to understand how complex language could arise,' he says.

'Again, not by programming it, but by finding out what are the fundamental mechanisms that allow the emergence of language by interaction with others and interaction with the environment.'

Steels says that his research demonstrates that forms of language between robots can emerge through their interaction with the environment and with others.

'The social elements, in the sense of interacting with others, are absolutely crucial for the emergence of complexity in the conceptual domain and in the language domain.'

Above all, Steels believes embodiment is crucial to this process, and it's why he uses real robots in his research.

'You could do a lot with simulation … but actually the real challenge is to show that language is emerging out of embodied interactions between these agents,' says the researcher.

Steels points to space as an example of why embodiment is so important: the concepts of left and right and front and back only make sense when you have a real body and matching orientation.

'The sensory inputs that you get from your vision system and from the interaction with the environment are the foundation on which you build concepts. These concepts and therefore the language have to be grounded in the bodily interaction with the world,' says Steels.

In relation to the idea of a mind upload, the obvious question is whether it's possible to be ourselves if we are not incarnated.

'Can you imagine still being yourself if you abandoned completely all the [physical] properties and qualities that characterise you as yourself?' asks Max Cappuccio.

For his part, Luc Steels reckons he's onto the big one— just how it is that we come to be ourselves. In the process he's most likely proving the power of REC as a coherent explanation of what and where mind is.

'I don't tell this to the philosophers, but I'm attacking the same problems that they have been tackling, but doing it with another methodology.'