What is integrated information theory, and why is it contested?

Integrated information theory, or IIT, is a mathematical theory of consciousness proposed by the neuroscientist Giulio Tononi in 2004. It holds that consciousness is identical to a certain kind of integrated information in a physical system, quantified by a measure called phi. It is controversial because in 2023 more than 100 researchers signed a letter calling it pseudoscience, and others pushed back on that charge. [1] [2] [3]

IIT aims to say when any system, whether a brain or some other lump of matter, is conscious, and to what degree. According to the Internet Encyclopedia of Philosophy, it requires a grouping of elements with physical cause-effect power over one another. That implies that only architecture with feedback loops, neural or computational, could realize consciousness. Of the groupings in a system, the one with the maximum of integrated information is identified with consciousness. [1] [4]

The theory also tries to preserve the idea that experience is immediate and unified, which its proponents say rules out functionalist and eliminativist accounts. Because it ties experience to measurable dynamics, the degree of consciousness is in principle measurable. It has also inspired clinical techniques for assessing consciousness in unresponsive patients. [1] [2]

The critics' position: a letter signed by 124 scholars argued that IIT should be labelled pseudoscience. The signatories worry that, although some aspects of the theory have been tested, the theory as a whole has not, so its bold and counter-intuitive implications lack experimental support. One such implication, pointed out by the computer scientist Scott Aaronson, is that an inactive grid of connected logic gates would be conscious. IIT also comes close to panpsychism, since it implies many more things are conscious than we ordinarily suppose. The critics fear that association with such a theory will damage the credibility of the whole field. [3] [4]

The respondents' position: several figures in the field called the letter poorly reasoned and disproportionate. They say the lack of experimental support is true of all current theories of consciousness and reflects limits of present neuroimaging. They fear that unsupported charges of pseudoscience will lead the entire science of consciousness to be seen that way. The theory has been championed by Christof Koch and treated respectfully by David Chalmers. A later survey of researchers found only a small minority fully endorsing the pseudoscience label, according to Wikipedia. [4] [5] [2]

The sources do not say that either side has prevailed. The dispute is about the whole theory's testability and the fairness of the label, not about whether parts of it make testable predictions. [4]

In 2020 a consortium secured $3.2 million from the Templeton World Charity Foundation for adversarial experiments, in which rival theorists design studies together and agree in advance which results favour which theory. In June 2023 the first results pitted IIT against global workspace theory, which says information becomes conscious when it is broadcast to many systems across the brain. The results were mixed. Some confirmed parts of IIT and some parts of global workspace theory, and on balance there was arguably a slight advantage to IIT. The Atlantic reports that current technology is too crude to test either theory's most meaningful hypotheses. [6] [4]

One underlying difficulty, noted in The Conversation, is that consciousness is known privately rather than as a publicly observable phenomenon, which makes experimental adjudication hard. [4]