Consciousness as depiction

It has become a commonplace for reviews of books on consciousness to ask: Do we really need another book on consciousness? But in reviewing Igor Aleksander's latest book, I can brush this question aside, as there are relatively few books on its particular sub-topic: machine consciousness (other notable exceptions include, e.g. [1,2]). 'Machine consciousness' is perhaps a misleading term – it conjures up clunky robots from 1950s B-movies. The term 'artificial consciousness' might do just as well if not better, and this is particularly so in discussing Aleksander's work. Not only is it concerned with the engineering task of creating artificial systems that either model experience or are themselves truly aware; it is also concerned with the scientific task of illuminating natural consciousness by means of such work, a synthetic component that is often lacking in books on consciousness. Although Aleksander has been investigating related issues for many years, his work is part of a recent resurgence of interest in machine consciousness, evidence for which can be found in the steady flow meetings on the More specifically, Aleksander's work is an example of an approach that is currently finding increasing support in the machine consciousness community: the imagination or simulation approach (e.g. [3,4]). On this view, a key aspect of consciousness is the ability of an agent, say a robot, to 'imagine' (represent in a sensory-motor-grounded manner) experiences it is not currently having (e.g. the ability to answer the question 'what would I see if I turned my camera this way?'). The robot can then use such expectations, in conjunction with some basic affective system ('bumper sensors being activated is bad') to assist it in deciding what to do next ('I shouldn't reverse, because if I did, my bumper sensors would be activated'). Aleksander's book is not a summary or analysis of the different work being done in this sub-sub-field, nor was it intended to be. But it does set out to do three things that need doing: (1) clarify the approach, giving specific mechanisms when possible; (2) show how the approach can be used to model or explain various features of consciousness; (3) defend the approach against some standard philosophical objections. A specific form of the first problem that all workers in this field must confront is: What makes this approach different from familiar AI models of mind? Most if not all traditional AI systems are capable of representing …