Brain scans taken as people slip into anaesthesia are offering support for one of the foremost explanations of consciousness. The approach may lead to progress in understanding the brain as well as new ways to test awareness in people with medical conditions of consciousness, such as those in a vegetative state after head injuries.
快猫短视频s and philosophers have long struggled to explain how the brain produces consciousness 鈥� the feeling of being aware of your surroundings or internal sensations and thoughts 鈥� and there are a number of competing ideas.
One called integrated information theory (IIT), first proposed in 2004 by at the University of Wisconsin-Madison, says something has a higher level of consciousness if the interactions between its components yield more information than when reduced to just its components. In other words, the whole is greater than the sum of its parts. The concept can quantify the complexity of any information-processing system, from brains to computers, but is a work in progress.
Advertisement
IIT predicts that it is possible to calculate a mathematical value for the level of consciousness, termed phi, of any information-processing system with known structure and functioning. But this requires a lot of calculations and as the number of connection points, or nodes, within an information-processing network grows, the maths involved rapidly gets exponentially bigger.
Currently, phi can only be calculated for systems with fewer than about 10 nodes. So if each of the roughly 86 billion neurons in the human brain is treated as one node, it would be impossible to work out its value of phi, even if we knew its complete structure.
To get round that problem, at the University of Western Ontario, Canada, and his team used brain-scanning data to calculate phi for simplified models of specific neural networks within the human brain that have known functions, such as the visual cortex.
Instead of regarding each neuron as a node, they treated small, anatomically defined brain areas as nodes, selecting five nodes for each network, doing this for 11 networks in total.
To get the data, the team asked 17 people to lie in a brain scanner as they experienced four different states of consciousness 鈥� awake, mildly sedated, unconscious and in a recovery stage 鈥� brought about by using anaesthetic.
The scans were made using functional magnetic resonance imaging (fMRI). This allowed the team to assign a value of 1 (relatively high activity) or 0 (low activity) to each small brain area, over successive time points.
To calculate phi for each network, a huge data set made from the scans was fed into a software package written by Tononi and his team. The results from all 17 people in the study had to be merged to have enough data without requiring them to spend too long in the scanner.
For two brain networks, the researchers found that phi was lower when people were in deep anaesthesia than while awake, with an intermediate level for mild anaesthesia. The networks 鈥� the frontoparietal and dorsal attention networks 鈥� were already thought to play a key role in consciousness. The other networks didn鈥檛 show a clear, decreasing trend in phi from an awake state through mild anaesthesia to deep anaesthesia, but showed small movements of phi either rising or falling.
As these two networks were the only ones where the calculated value for phi behaved as expected, this lends further support to their role in consciousness, says Soddu.
This is the first time phi has been calculated from brain-scanning data using the current incarnation of IIT, known as version 3.0, although similar studies have been done using the previous version, called 2.0.
Soddu鈥檚 team is currently investigating if calculations of phi for these brain networks could be used to indicate brain function in people with medical conditions of consciousness. While those results aren鈥檛 yet published, 鈥渢hey go in the right direction鈥�, he says.
at the University of Michigan Medical School in Ann Arbor says it would have been better if phi had been calculated for individuals.
at the University of Wisconsin-Madison, who helped develop IIT with Tononi but wasn鈥檛 involved in this research, says the approach only gives an approximation of phi for the human brain. But she, Tononi and others have now further refined IIT and version 4.0 will be published shortly, she says.
Yet the work may not convince critics of IIT, such as at the University of Texas at Austin. 鈥淚f you look at brains in fMRI while they鈥檙e being sedated and woken up, it鈥檚 clear and uncontroversial that you can see large changes in聽activity levels and in patterns of聽information transfer between different regions,鈥� he says. 鈥淵ou can surely learn many interesting things by examining those patterns, with or without IIT.鈥�
Communications Biology
