Shashwat's Reviews > The Feeling of Life Itself: Why Consciousness Is Widespread but Can't Be Computed

“Experience is found in unexpected places, in animals both large and small, but not in computers though they speak in tongues.”

I may have overreached with this book. Though Christof Koch brilliantly tries to simplify the complex mind-body problem, the problem itself requires one to be well-versed in multiple fields. Anyway, let me try my best to review what I managed to understand.

Consciousness is, simply, the feeling of life itself. According to the Integrated Information Theory (IIT), which is the leading theory on Consciousness today, Consciousness is definite, informative, integrated, structured, and (most importantly) exists for itself – or has causal effect on itself.
The theory goes on to describe how the neurology of consciousness has allowed scientists to develop consciousness meters based on EEG principles that work well in clinical settings. Most animals, perhaps all, experience life in some form or another, and are therefore conscious. However, he also argues that computers and AI don’t have the same mechanism for consciousness and may never have. Even a perfect software model of the brain may never be conscious, because consciousness is not a type of computation, or a clever hack. It is being itself.

IIT posits that any one conscious experience is IDENTICAL to a maximally irreducible cause-effect structure in the brain and is determined by the causal properties of any physical system acting upon itself.

Multiple feedback loops are essential for an inner experience. Computers and AI work on feedforward circuitry which, though may have many processing layers, is reducible to its components. They do not have sustained feedback to have causal powers over themselves. Conventional digital computers, built out of circuit components (dry hardware) with sparse connectivity and little overlap among their inputs and their outputs do not constitute a Whole. Yes, it may be possible to build computing machinery that closely mimics neural architecture, but we are still far from it.