#giulio tononi

[aioleis]

Integrated Information Theory Φ by Giulio Tononi

Integrated Information Theory (IIT) is a framework for understanding consciousness that has been developed by neuroscientist Giulio Tononi and further explored by collaborators such as Melanie Boly, Marcello Massimini, and Christof Koch. The theory proposes a quantitative measure, denoted as Φ (phi), which is intended to indicate the level of integrated information within a system—a measure that, according to IIT, correlates directly with consciousness.

Key Concepts of Integrated Information Theory

Integrated Information (Φ): IIT posits that consciousness arises from the ability of a system to integrate information in a way that the whole is more than the sum of its parts. This integration is measured by Φ, which quantifies how much more information is generated by the interconnected system as a whole compared to what would be generated by its individual parts working independently.

Intrinsic Cause-Effect Power: The theory argues that the physical substrate of consciousness must possess a high degree of intrinsic cause-effect power. This means that the system must be able to determine its own state through a complex of internal interactions that cannot be reduced merely to the behavior of its components.

Necessary and Sufficient Conditions for Consciousness: IIT proposes that a high Φ is both necessary and sufficient for consciousness. This suggests that any system with a sufficiently high Φ, regardless of its makeup (biological or artificial), should possess consciousness.

Counterintuitive Predictions: IIT leads to some predictions that may seem counterintuitive. For example, a system can have high Φ and thus be considered conscious even if it does not interact with its environment or display outward behavior that typically accompanies consciousness in humans and other animals.

Applications and Implications

Assessment of Consciousness: IIT provides a theoretical basis for developing tools to assess consciousness, particularly in cases where traditional behavioral indicators are absent, such as in non-communicative patients or potentially in machines.

Philosophical and Ethical Questions: The theory raises significant questions about the nature of consciousness, the ethical treatment of systems (biological or artificial) that might have high Φ, and the understanding of consciousness in medical and legal contexts.

Criticisms and Challenges Disconnection from Behavior: One of the main criticisms of IIT is that it divorces consciousness from observable behavior. Traditional views often link consciousness closely with behavioral responses, and IIT's framework challenges these assumptions by suggesting that consciousness can exist even in isolated systems that do not interact with their surroundings.

Measurability of Φ: Calculating Φ in complex systems is technically challenging and computationally intensive. This raises practical questions about the feasibility of applying IIT in real-world scenarios.

Philosophical Underpinnings: The assumption that consciousness can be fully explained as a form of information integration is philosophically controversial and not universally accepted.

***

Abstract

In this Opinion article, we discuss how integrated information theory accounts for several aspects of the relationship between consciousness and the brain. Integrated information theory starts from the essential properties of phenomenal experience, from which it derives the requirements for the physical substrate of consciousness. It argues that the physical substrate of consciousness must be a maximum of intrinsic cause–effect power and provides a means to determine, in principle, the quality and quantity of experience. The theory leads to some counterintuitive predictions and can be used to develop new tools for assessing consciousness in non-communicative patients.

Article >

***

"The neuroscientist Giulio Tononi has purported to supply just such a measure, named Φ, within the rubric of so-called ‘integrated information theory’.

Here, Φ describes the extent to which a system is, in a specific information-theoretic sense, more than the sum of its parts.

For Tononi, consciousness is Φ in much the same sense that water is H2O.

So, integrated information theory claims to supply both necessary and sufficient conditions for the presence of consciousness in any given dynamical system.

The chief difficulty with this approach is that it divorces consciousness from behaviour.

A completely self-contained system can have high Φ despite having no interactions with anything outside itself.

Yet our everyday concept of consciousness is inherently bound up with behaviour. If you remark to me that someone was or was not aware of something (an oncoming car, say, or a friend passing in the corridor) it gives me certain expectations about their behaviour (they will or won’t brake, they will or won’t say hello).

I might make similar remarks to you about what I was aware of in order to account for my own behaviour. ‘I can hardly tell the difference between those two colours’; ‘I’m trying to work out that sum in my head, but it’s too hard’; ‘I’ve just remembered what she said’; ‘It doesn’t hurt as much now’ – all these sentences help to explain my behaviour to fellow speakers of my language and play a role in our everyday social activity.

They help us keep each other informed about what we have done in the past, are doing right now or are likely to do in the future. Problematic — What if I say to you, there is no past, there is no future ? Everything is in exist in me alpha&omega.

It’s only when we do philosophy that we start to speak of consciousness, experience and sensation in terms of private subjectivity.

[maaarine]

Are we close to solving the puzzle of consciousness? (David Robson, BBC Future, Mar 27 2019)

“Tononi proposes that any conscious experience needs to be structured, for instance – if you look at the space around you, you can distinguish the position of objects relative to each other.

It’s also specific and "differentiated" – each experience will be different depending on the particular circumstances, meaning there are a huge number of possible experiences. 

And it is integrated. If you look at a red book on a table, its shape and colour and location – although initially processed separately in the brain – are all held together at once in a single conscious experience. 

We even combine information from many different senses – what Virginia Woolf described as the “incessant shower of innumerable atoms” – into a single sense of the here and now.

From these axioms, Tononi proposes that we can identify a person’s (or an animal’s, or even a computer’s) consciousness from the level of “information integration” that is possible in the brain (or CPU). 

According to his theory, the more information that is shared and processed between many different components to contribute to that single experience, then the higher the level of consciousness. 

Perhaps the best way to understand what this means in practice is to compare the brain’s visual system to a digital camera.

A camera captures the light hitting each pixel of the image sensor – which is clearly a huge amount of total information.

But the pixels are not “talking” to each other or sharing information: each one is independently recording a tiny part of the scene.

And without that integration, it can’t have a rich conscious experience.

Like the digital camera, the human retina contains many sensors that initially capture small elements of the scene.

But that data is then shared and processed across many different brain regions. (…)

The same goes for our memories. Unlike a digital camera’s library of photos, we don’t store each experience separately. 

They are combined and cross-linked to form a meaningful narrative.

Every time we experience something new, it is integrated with that previous information. (…)

By altering the levels of important neurotransmitters, the drugs appeared to have “broken down” the brain’s information integration – and this corresponded to the participants’ complete lack of awareness during the experiment.

Their inner experience had faded to black."

[principleofplenitude]

Some researchers, such as the cognitive scientist Stanislas Dehaene of the Collège de France in Paris, suggest that this conscious behavior arises when we hold a piece of information in a “global workspace” within the brain, where it can be broadcast to brain modules associated with specific tasks. This workspace, he says, imposes a kind of information bottleneck: Only when the first conscious notion slips away can another take its place. According to Dehaene, brain-imaging studies suggest this “conscious bottleneck” is a distributed network of neurons in the brain’s prefrontal cortex. This picture of consciousness is called global workspace theory (GWT). In this view, consciousness is created by the workspace itself — and so it should be a feature of any information-processing system capable of broadcasting information to other processing centers. It makes consciousness a kind of computation for motivating and guiding actions. “Once you have information and the information is made broadly available, in that act consciousness occurs,” said Christof Koch, chief scientist and president of the Allen Institute for Brain Science in Seattle. He has been developing a competing theory in collaboration with its originator, the neuroscientist Giulio Tononi of the University of Wisconsin-Madison. They say that consciousness is not something that arises while turning inputs into outputs but rather an intrinsic property of the right kind of cognitive network, one that has specific features in its architecture. Tononi christened this view integrated information theory (IIT). In contrast to GWT, which starts by asking what the brain does to create the conscious experience, IIT begins instead with the experience. “To be conscious is to have an experience,” Tononi said. It doesn’t have to be an experience about anything, although it can be; dreams, or some “blank mind” states attained by meditation also count as conscious experiences. Tononi has sought to identify the essential features of these experiences: namely, that they are subjective (they exist only for the conscious entity), structured (their contents relate to one another: “the blue book is on the table”), specific (the book is blue, not red), unified (there is only one experience at a time) and definitive (there are bounds to what the experience contains). From these axioms, Tononi and Koch claim to have deduced the properties that a physical system must possess if it is to have some degree of consciousness.

“Neuroscience Readies for a Showdown Over Consciousness Ideas“ from Quanta Magazine

[maaarine]

MBTI & Science Giulio Tononi: INTJ

“Giulio Tononi is a neuroscientist and psychiatrist who holds the David P. White Chair in Sleep Medicine, as well as a Distinguished Chair in Consciousness Science, at the University of Wisconsin. (…)

Tononi is also a leader in the field of consciousness studies, and has co-authored a book on the subject with Nobel prize winner Gerald Edelman.

He developed the integrated information theory (IIT): a scientific theory of what consciousness is, how it can be measured, how it is correlated with brain states, and why it fades when we fall into dreamless sleep and returns when we dream. 

The theory is being tested with neuroimaging, TMS, and computer models. His work has been described as "the only really promising fundamental theory of consciousness" by collaborator Christof Koch.”

Sources: video, wiki/Giulio_Tononi

[vonneumannmachine]

No, ni LLMs son conscientes ni mi smartphone tiene hambre

Llevo unos días un tanto consternado porque, a pesar de que estoy completamente seguro que mi blog es leído fervientemente en Silicon Valley, ahora se están duplicando los trabajos y publicaciones que hablan de consciencia en grandes modelos de lenguaje con desenfadada frescura y sin demasiado pudor. No obstante, también me da cierta alegría que el campo de la machine consciousness, que lleva…

View On WordPress

[averycanadianfilm]

You might recall that last week I wrote a post criticizing Integrated Information Theory (IIT), and its apparent implication that a simple Reed-Solomon decoding circuit would, if scaled to a large enough size, bring into being a consciousness vastly exceeding our own. On Wednesday Giulio Tononi, the creator of IIT, was kind enough to send me a fascinating 14-page rebuttal, and to give me permission to share it here: Why Scott should stare at a blank wall and reconsider (or, the conscious grid) If you’re interested in this subject at all, then I strongly recommend reading Giulio’s response before continuing further. But for those who want the tl;dr: Giulio, not one to battle strawmen, first restates my own argument against IIT with crystal clarity. And while he has some minor quibbles (e.g., apparently my calculations of Φ didn’t use the most recent, “3.0” version of IIT), he wisely sets those aside in order to focus on the core question: according to IIT, are all sorts of simple expander graphs conscious?

Scott Aaronson’s debate with Giulio Tononi May 2014

[timespanner]

Are we close to solving the puzzle of consciousness?

Giulio Tononi’s "integrated information theory" might solve neuroscience’s biggest puzzle

[the-apple-core]

...it would seem that some people cannot help but write the same story all their life.

"Preface," Phi: A Voyage from the Brain to the Soul