Statement :
Dennett's Predictive Bayesian Coding Theory
In the early 90's the distinguished philosopher Daniel Dennett was very involved in the consciousness debate. In addition to many published papers, his "Multiple Drafts" theory was described in his book: "Consciousness Explained". Then, for a time, he "lost interest in the topic." Some people assumed, from his silence, that he was conceding to dualistic theories of consciousness. However, in 2012, at the "Evolution and Function of Consciousness" conference he presented a talk where he talked about "The phenomenal access consciousness distinction". (
Talk on YouTube).
He pointed out that his critics didn't lose interest, that his loss of interest was possibly a "tactical mistake" (6:00) and that he was now interested in returning to the debate. In that talk he summarized an updated version of his theory documented here.
In that talk he points out that he understands and admits to the powerful arguments for dualism (12:00) but exhorts us to resist them. He demonstrates and focuses his discussion around an after image of red which is obviously not on the screen, nor on the retina. So in this talk he asks and focuses on what the emerging expert consensus agrees is the most important part of consciousness, and the focus of this survey topic: What, Where and How, is this redness after image. He coined the term "Quining Qualia" where he argues that even though we know it exists, we should ignore it.
Dennett argues that Ned Block has inverted reasoning and that "He thinks 'phenomenal' consciousness is the causal bases of access consciousness while in fact it is an effect of access consciousness, not a cause!" (1:05:45) And ultimately predicts that this effect emerges via Predictive Bayesian Coding. Dennett points out, in that video, that qualia are achieved via the predictive powers of the bayes algorithm. He points out that he doesn't have time to explain it in more detail, in the video, and provides multiple references to find out more. And of course, you can google for the term "predictive bayesian coding" or bayes theorem.