The principal thesis in Bringsjord's book, What Robots Can and Can't Be (1992), consists in the assertion that, although destined to excel at Turing tests, robots will never become people. However, the book provides a comprehensive review of many computer theoretical and logico-mathematical aspects related to automata and control, while little is discussed about what is, to the author of the present review, the most important difference between persons and machines: the incorporation of consciousness in its plenitude, including its sensation-related aspects. Remarks on some consciousness-related aspects that have been overlooked as well as how they may contribute to constructing a more realistic picture of the mechanical aspects of consciousness are also included.
"I may not be able to define one [a robot], but I know when I see one." [Attributed to Joseph Engelberger, "father of the industrial robot", in Schodt (1988).]
1. Bringsjord's book What Robots Can and Can't Be (1992) possesses an self-explanatory title. More specifically, it is concerned with how the person-building project will fare as humanity steps into the next century and beyond (hopefully). The key thesis is that, in spite of passing Turing tests with flying colors, robots will never become people. Although such a proposal caught my attention from the beginning for the implications and relationships it bears with consciousness, it turned out that the book concentrates too heavily on the logico-mathematical and computer theoretical aspects of the problem, paying little attention to the issue that I consider to be essential, namely the experiencing (qualia) aspect of consciousness. (A nice discussion of the relationship between psychology and mathematics can be found in Julesz (1995)). Despite my high consideration for mathematicians and logicians, it seems to me that we currently know too little about the nature and organization of the brain's workings, and even less about subjective experience, to enable us to embark on a completely formal and mathematical approach to one of the most challenging and ultimate problems in science. Except for that, I thoroughly enjoyed Bringsjord's contribution for its clear mood, many interesting insights and the comprehensive and updated review and discussion of the many trends and developments in AI, computational theory and logico-philosophy. Issues ranging from the various Turing tests that have been proposed to the Goedel incompleteness theorem, free will and introspection have all been included and commented upon. The book also presents some clever insights from the mathematical and computer theoretical perspectives.
2. This review discusses Bringsjord's principal thesis and the way it is supported, followed by an argument that such an approach misses the main question that is attracting so much scientific attention at the moment. The last section provides a personal discussion of some consciousness-related aspects that have been deserving greater attention and which could help us to construct a more comprehensive picture of the mechanical aspects of consciousness.
3. As stated above, Bringsjord's book is a stubborn (in the good sense) defense of the thesis that robots will never be persons. Although I do not intend to discuss in detail all the formalizations used to defend such a thesis, one of the main arguments presented in support of this hypothesis consists of the assumption that persons are not automata. Since this is a prerequisite for the Person Building Project, it follows that such a project is doomed to failure. To quote the author, this is, in a nutshell, the basic rationale of the book.
4. Although some attention has been drawn to the important issue of personhood characterization (e.g., pp. 82-85), the important fact that this is a fuzzy and relative concept does not seem to have received the attention it deserves. More specifically, persons have been characterized by Bringsjord according to the following four propositions (pp. 83-84):
P1 - Persons are genuine individual things, not logical constructions;
P2 - Persons are bearers of psychological, Cartesian, or self-
representing properties (such as fearing unicorns, believing that quantum mechanics is silly, etc.), and when a person has such a property she is said to be in a mental state;
P3 - Some persons bring about some state of affairs;
P4 - There are persons.
5. Bringsjord anticipated he would probably be attacked by those who do not believe in personhood. Yet, in the present review he will be challenged not for that reason - I do believe I am a person - but because it seems clear to me that there is more to personhood than computer theory and logico-philosophy. Getting directly to the point, I believe one of the most fundamental differences between machines and humans is that the latter possess consciousness in its full plenitude. Now, I mean consciousness in the sense that Chalmers (1993, in press) has so aptly characterized, with emphasis on his identification of two principal aspects of consciousness: (i) the "mechanical" (my denomination), that includes the awareness of oneself as well as one's position to the world, and (ii) the experiencing aspect, that is the feelings one has while listening to one of Duparc's songs or when one realizes it is Monday. As expressed in Costa (1994), I am a believer that while aspect (i) will be fully incorporated into machines, the achievement of (ii) will be a daunting task at the very least. According to such hypotheses, the question whether robots will become persons can be restated as whether robots will ever incorporate the sensation aspect of consciousness. Unfortunately, the strictly personal and subjective nature of sensation (I feel, thus I am) makes the definitive answer to such a question beyond the scope of current science. (See Chalmers (1993, in press) for a brilliant treatment of such perspectives).
6. Of course, consciousness is a condition that is only necessary and not sufficient for robots becoming persons. Another requisite that has not been fully explored by Bringsjord is the developmental aspect of consciousness, that is the history of the individual psychological development. It is notable that one of the strategies used by the creators of the female automaton Rachel in the film Blade Runner to make her feel she is a human is the production of photographs of her supposed earlier familiar life.
7. All in all, these are the reasons why I think Bringsjord's approach may have missed the key issue of the question it proposes to answer. In the remaining section I will briefly discuss some consciousness- related issues that, though largely overlooked by recent studies, present potential answers to parts of the riddles that plague the above proposed hypotheses.
8. It can be stated without exaggerating too much that there are as many theories of consciousness nowadays as there are researchers interested in that tantalizing issue. Although consciousness research was until recently largely a speculative endeavour, where the literary skills of the proponents often played a fundamental role, the identification of sensation (or qualia) as the really tricky issue (Chalmers, 1993) is perhaps one of the greatest and few substantial breakthroughs that has been achieved in the study of consciousness. And for one important reason: It has separated the aspects of consciousness that can be properly treated by current science from those that will have to await more substantial advances. That such an almost obvious fact, and one that has not yet become a consensus, has taken such a long time to be identified may be a poignant indication of our inherent naivety towards consciousness analysis (the fish in the water). This section provides a personal view of some neglected consciousness-related aspects that may help to clarify some of the above discussed issues.
9. First, I would like to observe that the separation of the consciousness properties poses the following interesting question: Are the mechanical and sensational aspects of consciousness reciprocally dependent? In case they are not, would it be possible that sensation is a property shared by other animals and even machines? If that is so, would human consciousness be no more than a refined version that was achieved through the comprehensive integration of feelings and self and world awareness? Unfortunately, such questions can not be scientifically attacked at the moment because of the sensation aspect of consciousness that is involved. Nevertheless, such proposals lead to a second interesting possibility, namely that consciousness is a graded and not a binary (all-or-nothing) property. In other words, there is no clear line dividing conscious human beings from those who are not. For instance, in which stage during the development of a progressive brain disease such as Alzheimer's, or even during the process of normal aging, can one individual be said to be no longer conscious? Moreover, conditions in which people are raised, especially the cultural, educational and religious aspects of the environment in which they are born, are known to have definite effects on their consciousness. A ubiquitous illustration of such a fact is the more cosmic characteristic of Hinduism (cosmic consciousness) contrasted with the more individual tendency of the Jewish-Christian tradition. Can we say people from these two distinct traditions share the same kind of consciousness?
10. The third and last issue I would like to address in this section concerns the influence vision may have had over the epigenesy and ontogenesy of consciousness. It is an acknowledged fact that we are creatures controlled by (hedonic) sensations which continuously motivate us to strive for food and other basic needs, whose fulfillment is indicated to us also through sensations. While other creatures have managed to survive by relying on other senses, we humans are essentially visual creatures. The proposal I would like to advance here is that, being the most important sense for humans, vision may have had a much more decisive influence in shaping the mechanical (and perhaps also the sensational) aspect of human consciousness than is commonly acknowledged. For one thing, vision is virtually indispensable to our mastering the sensory-motor integration that is so fundamental for the development of the awareness of ourselves and the world. For instance, I find it extremely difficult not to think of our progressive visual realization that the ten cylinder-like objects before us are under our control as not having a primordial importance for the development of consciousness. Moreover, in addition to such reflexive interactions, it is quite reasonable to suppose that, as a consequence of its importance, vision may have to a large extent shaped the principal processes in our brain, both from the anatomic and functional perspectives. That this has been so in the lower hierarchy visual areas is an established fact expressed by the topographic organization of the many visual regions in Broadmann's areas 17 and 18. Further, there are some remarkable relationships between the way in which we see the world and the way we think about it. Take for instance active foveal vision: It is a remarkable coincidence that the sequential, episodic flow of our thoughts reminds us so closely of the saccadic movements that are needed for visual analysis. What is more, the temporal association integration mechanism that allows us to make sense of a whole science from its constituent pieces seems to bear a striking analogy to the associative flavour of our sequence of thoughts.
11. It should be observed that such hypotheses are not necessarily contrary to Bringsjord's developments. On the contrary, it may be only through the incorporation of insights from as many areas as anatomy, philosophy, mathematics, physiology and physics, to name but a few, that we will in the end succeed in our quest for the nature of human and robotic consciousness.
12. This text has provided a strictly personal critical review of Bringsjord's What Robots Can and Can't Be. Although this book provides a nice review and discussion of the main paradigms in AI and logico-philosophy, I believe it has missed the big issue behind robots becoming or not becoming persons, namely the incorporation of full blown consciousness in both its mechanical and experiencing flavours. It also seems to have overlooked some other important related issues such as the fundamental role the psychological development has over personhood. In addition to discussing such issues according to the perspective on consciousness recently identified by Chalmers (1993, in press), the present work has also identified some consciousness- related issues, such as the graded nature of consciousness and the substantial influence vision may have had upon the development of consciousness, that are deserving of further attention for the potential they may represent, when taken integrately with other perspectives and evidences, for clarifying some of the secrets of consciousness.
Bringsjord, S. (1992) What Robots Can and Can't Be. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Bringsjord, S. (1994) Precis of: What Robots Can and Can't Be. PSYCOLOQUY 5(59) robot-consciousness.1.bringsjord.
Costa, L. da F. (1994) Getting the Ghost Out of the Machine. PSYCHE, 1039-723X/93.
Chalmers, D.J. (in press). Facing up to the problem of consciousness. In Proceedings: Toward a scientific basis for consciousness. Cambridge, Mass.: MIT Press.
Chalmers, D.J. (1993). Towards a Theory of Consciousness, PhD Thesis, Indiana University, USA.
Julesz, B. (1995). Dialogues on Perception. Cambridge, Mass.: MIT Press.
Schodt, F.L. (1988). Inside the Robot Kingdom. Tokyo, Kodansha International.