Consciousness and Robot Sentience (Series on Machine Consciousness) [Hardcover]

This book purports to show how, theoretically and practically, robot sentience is possible. Haikonen's approach combines his answer to the philosophical question of what constitutes consciousness, with his own rather ingenious "cognitive architecture" for building robots that might well satisfy the criteria for what we call "being conscious".

The author is actually rather confident, that the problem of giving a reasonable definition or description of what counts as consciousness, has been solved. Readers can draw their own conclusions on that one - personally I think Haikonen neatly describes the criteria for calling something conscious, and also provides nice system architecture for realising the kind of "machine-qualia" that robots might end up with - I think he ends up with a workable operational definition of sentience, which his own Cognitive Architecture is consistent with. Anyway, the main positives I take from this book are:

1. A neat associative neural-type architecture, which manages among other things to generate symbolic processing from sub-symbolic processing.

2. A workable mechanism for generating the diaphanous perceptual experience of both direct perception and e.g. imagined and remembered experience: a perception/response feedback loop that returns the internal neural activities into virtual sensory percepts. Although I don't think that the existence of a mechanism or five solves all philosophical and phenomenological issues surrounding perception, on the other hand Haikonen provides a lot more than the typical philosopher would provide (philosophers will generally just wring their hands about the "hard problem" of the diaphanousness of inner experience, while mostly ignoring how such experience could even come about).

3. Haikonen, unlike most philosophers of mind, has built a conscious robot!! (The author himself does not make maybe such an extravagant claim (personal correspondence), but I think that if, as seems likely to me, the Haikonen Cognitive Architecture and supporting philosophical arguments constitute a viable enterprise, then his robot meets the goals of this enterprise). You can find XCR-1 on YouTube, it's quite smart as machines go - but the main thing here that impresses me is that he built the robot using the Haikonen Cognitive Architecture design principles - and so he has a "living" proof of the utility and plausibility of the Haikonen approach to consciousness. And that is much, much more than most theorists of mind have ever contributed.

Overall: this is a book that has a clear style of exposition, yet contains a lot of nice little surprise factoids and diversions into unusual side-topics. A book which will teach philosophers new things about robots and cognitive architectures, but also an excellent primer for engineers grappling gamely with the issue of how to find a match between traditional engineering approaches to AI-design juxtaposed against the stubbornly eclectic design of our very own biological machines with their inner experiences.