This is Part 2 of a series on determinism, free will, and consciousness. If you haven't read the Part 1 where we explore the idea of determinism and free will you should definitely check it out. I mean it's written in dot-points after all:
In this post, we're going to look at a new theory called Integrated Information Theory. It's pretty dense, but after we get through this we are going to tie it all together in Part 3.
What is Integrated Information Theory?
I'm going to try summarise as succinctly as possible, what this exciting new(ish) theory is all about:
- Integrated Information Theory (IIT from here on out) is a theory of consciousness.
- Proposed by neuroscientist Giulio Tononi in 2004.
- It tackles the "hard problem" from the alternate route. Rather than starting with neural activity in the brain and theorising how you get to consciousness experience, it starts with consciousness as a given, and works backwards.
- This is similar to how phenomenon such as gravity are explained. We still are unable to describe how you get from matter to gravitational attraction, but by accepting that matter does attract, we can create a theory of gravity to account for and predict it's behaviour.
IIT states that every information system which is connected and differentiated is conscious to some degree.
- In other words, consciousness is what it feels like to be an information system.
Give me the details
IIT defines an information system as:
- "a set of elements, each with two or more internal states, inputs that influence that state, and outputs that are influenced by that state" - Wiki
- So basically, you need to have multiple things with a state (on/off) that are connected together and influence each other.
- Some examples would be: transistors inside a computer, the neural network that is your brain, or even the neurons lining your gut!
IIT observes five axioms (properties) that seem to be essential to consciousness.
- Intrinsic existence – each conscious experience is intrinsically real; this means it must have cause-and-effect power over itself, otherwise it might well not exist.
- Composition – consciousness is structured in that it consists of multiple distinct components. This conscious experience right now consists of you reading at your computer, in your underpants, with your jaw slightly open in mind-boggling amazement.
- Information – each moment of consciousness is specific (differentiated). In a system, information is the structure that constrains the past, present, and future in a specific way. These are known as cause-effect repertoires (rulesets), and they are what make the experience of sadness different from the experience of happiness.
- Integration – the experience of consciousness is irreducible to its parts. This conscious experience cannot be split up into multiple experiences. This moment cannot be split into separate experiences of you reading this article, you in your underpants, and your jaw slightly open. The experience is the unification of all these together.
- Exclusion – each distinct moment of consciousness excludes all other moments. You cannot have two experiences at the same time (the union would be a different experience altogether.)
Confused? It's ok. Basically, the main thing to take away is:
Any self-influencing system that integrates information in a way that the result is greater than the sum of it's parts is conscious.
Other things you should know about IIT:
- Phi (ɸ) is the term that mathematically represents the degree to which a system is connected and differentiated. Aka, Phi is the correlate of consciousness, the higher Phi, the more conscious the system is.
- A quale (plural: qualia) is what it feels like to experience something. The colour red has a particular quale, as does happiness, sadness, etc. To be more exact, each moment of happiness, sadness, redness combined with all the other inputs to your brain, combine to an extremely complex quale.
- A qualia shape is the term given to represent what it feels like to be an information system at any given moment in time. This "shape" is not dependant on the underlying structure (neural activity) but a figurative mathematical representation of the system at any given time.
- Consciousness is computationally irreducibility - this means that there is no "short-cut" to predicting the output of the system, no model that can be used. When a system is computationally irreducible, then the only way to predict it's behaviour accurately is to replicate it. You'll see why this is important later.
Practical applications of IIT
Alright, so you kinda get the idea behind this radical theory. What can it do for us in terms of explanation and prediction (the essence of any scientific theory):
- Determining PVS - Persistent Vegetative State is the term for someone who shows no sign of consciousness. The current method for determining this is a set of crude rules and measurements. Instead with potential PVS patients we could use inductive currents and brain scanning to calculate the degree to which their neural activity is connected and differentiated to determine how conscious they are. This could help people make the difficult decisions in those tough times.
- Consciousness detecter - we could theoretically build a machine that pointed at anything, could calculate the level of Phi (how integrated and differentiated the system is) to determine what actually is conscious and what is not.
- Qualiascope - we could theoretically build a machine that pointed at anything could calculate what the qualia shape that the system is currently "feeling". This would allow us to compare the shapes of our own feelings, to find something that loosely matches the feeling.
- It changes everything - ok so the above things are mostly theoretical, but in my next post in the series I share my take on IIT and how it helped explain consciousness, free will, and ultimately life 🙌.
Other related ideas
Before I end this, here are a few other ideas that relate into IIT that you might want to dive deeper into:
- Complex systems - is any system where the observed behaviour is more than the sum of its parts. Emergence is used to describe behaviours/properties of the system that cannot be calculated by only observing a single part.
- Computational irreducibility - A system where no model can accurately predict the result, the only way to accurately predict the result is to "run" the system. The brain is one of those systems in that the only way to know the exact outcome of billions of interconnected neurons is to simulate billions of interconnected neurons with the exact same inputs.
- Chaotic systems - are any complex system in mathematics where a tiny change to the input parameter causes dramatic changes to the output. An example of a chaotic system is the double pendulum swing.
- Butterfly effect – is the term used to describe the property of complex chaotic systems, where a minor change to the input values can have a major impact on the result. The term comes from the idea that a butterfly flapping its wings will indirectly influence the path of a tornado thousands of miles away.
- Cellular automata - a way of modelling complex systems through replicating the interaction of individual units. This is a neat site exploring this.
Next up: The Answers to Everything
Cool... enough. I applaud you for getting this far in this three-part series. The good thing to know is that Part 3 is where it all gets interesting; it's the big payday for all your hard work so far.
Next up we'll explore my take on IIT and how it relates to free will, determinism, and ultimately the meaning of life.