Levels of Explanation and Hierarchies of Self-Consciousness: Response to Díaz

by José Luis Bermúdez 


I am grateful to José Luis Díaz for his sympathetic comments and helpful criticisms. In The Paradox of Self-Consciousness I stressed the importance of showing how full-fledged self-consciousness, as manifested in conceptual first-person thoughts and comprehending use of the first person pronoun, can be built up from more primitive cognitive and affective capacities. I took 'primitive' in two different ways and corresponding to these there are two different "constructive projects". One capacity can be more primitive than another in ontogenetic terms. That is to say, the more primitive a capacity is the earlier it is acquired in the normal course of human development. Corresponding to this understanding of 'primitive' there is the constructive project of showing how full-fledged self-consciousness emerges ontogenetically from the foundations provided by forms of self-consciousness that are present from the beginning of life. On the second understanding of 'primitive', one capacity is more primitive than another if and only if the first is logically prior to the second. To say that one capacity is logically prior to another is to say that the second can be analysed in terms of the first, but not vice versa. When primitiveness is taken in these terms, the constructive project becomes the project of showing how full-fledged self-consciousness can be analysed in terms of forms of self-consciousness that do not involve concept possession and mastery of the first-person pronoun.

In his comments Professor Díaz concentrates on the first of these projects - the project of showing how full-fledged self-consciousness meets what in the book I called the Acquisition Constraint:

    The Acquisition Constraint If a given psychological capacity is psychologically real then there must be an explanation of how it is possible for an individual in the normal course of human development to acquire that cognitive capacity.

Díaz is sympathetic to the general approach exemplified by taking the Acquisition Constraint seriously, but feels that there are three potential problems that any account along the general lines of the account I offer will have to overcome.

Problem 1: There is a danger of confusing levels of explanation. Self-consciousness is a phenomenon that arises at what Dennett and others have termed the personal level, that is to say, the level of conscious experience of the world and one's own agency within the world (see the essays in Bermúdez and Elton 2000 for discussion of the distinction between personal and subpersonal levels). Clearly, therefore, it cannot be identified with the various forms of self-specifying information that exist at the neural level, such as the information about the agent's own location that is coded within the cognitive map of the perceived environment believed to be located in the hippocampus. But how are we to distinguish genuine forms of primitive self-consciousness at the personal level from mere self-specifying information at the subpersonal level?                 

Problem 2: In thinking about self-consciousness I began with what I termed the paradox of self-consciousness. The paradox arises for theories that analyse self-consciousness in terms of the ability to think and enunciate first-person thoughts whose natural linguistic expression employs the first-person pronoun. Such theories are circular because self-consciousness is presupposed in linguistic mastery of the first-person pronoun. The constructive project of showing how full-fledged self-consciousness both emerges ontogenetically from, and can be analysed in terms of, the foundations provided by forms of self-consciousness that are present from the beginning of life was offered as a solution to the paradox. But, asks Díaz, does the paradox not reappear? Are not the most primitive forms of self-consciousness just as unanalysable and mysterious as full-fledged self-consciousness was initially found to be?

Problem 3: Both the analytic and the ontogenetic constructive projects depend upon the idea that there is a basic hierarchy of forms of forms of self-consciousness. At the bottom of the hierarchy I placed the self-specifying information that appears in the content of visual perception and the various different types of somatic proprioception. Díaz raises the question, however, of just how primitive these forms of self-consciousness are - and consequently of how stable the hierarchy of forms of self-consciousness should be taken to be. He points out that various aspects of the body-image are susceptible to top-down influence. Examples are to be found in the distortions of the body-image found in anorexics, obsessive bodybuilders and in patients suffering from anasognosia for hemiplegic neglect and other neuropsychological disorders.                

I will take these three problems in turn and try to show that the difficulties they present for my position are not insuperable.

PROBLEM 1 It is extremely important to distinguish personal and subpersonal levels of explanation, and Díaz is quite right to suggest that there is an ambiguity in terms like 'cognitive map' which can be taken in both a personal-level and a subpersonal-level way. How then are we to distinguish between genuine self-conscious thought at the personal level and mere self-specifying information at one or other level of subpersonal information?

The first step, of course, is to distinguish genuinely thinking behaviour from behaviour that is unthinking. Self-consciousness is a type of thinking and the approach I took in the book to identifying types of primitive or nonconceptual self-consciousness essentially involved finding experimental or ethological scenarios in which the behaviour of a non-linguistic creature seems to require explanation in terms of first-person thoughts. Clearly, there are types of behaviour that can only be explained by positing some form of subpersonal cognitive map on the part of the organism concerned, and yet which do not seem to qualify as genuinely thinking behaviours. A case in point would be the complex navigational calculations that are regularly performed subpersonally by migrating birds and many other organisms (Gallistel 1990). We do not need to attribute a personal-level cognitive map, and the first-person thoughts that go with it, in explaining why such creatures behave the way they do, because their behaviour lacks the flexibility and spontaneity that is required for the attribution of thoughts to be anything more than an anthropomorphic projection.

There seem to be five conditions that must be satisfied for mental representations to qualify as genuine thoughts:

    (1). They should serve to explain behaviour in situations where the connections between sensory input and behavioural output cannot be plotted in a lawlike manner.

    (2). They should admit of cognitive integration.

    (3). They should be compositionally structured in such a way that their elements can be constituents of other representational states.

    (4). They should have determinate contents.

    (5). They should permit the possibility of misrepresentation.

Once we have established that we are dealing with behaviour that can only be explained at the level of thought we have essentially moved beyond the sub-personal level. Of course, sub-personal facts will still be relevant to the explanations we give - not least as explaining what underlies and underpins the personal-level capacities that enable the creature to entertain the thoughts that we appeal to directly in our explanations. The answer to Díaz's question, therefore, is that we identify genuine forms of self-consciousness in virtue of the fact that they serve to explain thinking behaviour.

PROBLEM 2 The constructive response to the paradox of self-consciousness takes both ontogenetic and analytical forms. Both responses rely on the basic idea that there are certain primitive forms of self-consciousness that can serve as "building blocks" for the higher forms of self-consciousness. Díaz asks why the existence of such primitive building blocks should not itself be found puzzling and potentially paradoxical.

Ontogenetically speaking, the paradox of self-consciousness arises because neither full-fledged self-consciousness nor the linguistic ability to employ the first-person pronoun are innate, and yet each seems to presuppose the other. The paradox is essentially that it seems impossible to understand how we can "bootstrap" ourselves into full-fledged self-consciousness. Underlying this is the thought that mastery of the first person pronoun is something that must be learnt. Yet the thoughts that I need to grasp if I am to learn how to employ the first-person pronoun are not ones that are available to me before I have mastered the first-person pronoun. But this problem does not arise for such primitive forms of self-consciousness as the self-specifying information available in visual perception or in somatic proprioception. These can plausibly be viewed as innate. So issues of bootstrapping and learning do not apply.

Of course, there are genuine questions to be asked about how and when these forms of primitive self-consciousness emerged in evolutionary history. But such phylogenetic questions are fundamentally different from the questions that arise with regard to ontogeny. In discussing phylogeny one can appeal to mechanisms of random selection, genetic drift and so forth that are simply not available when one wants to explain an ontogenetic progression. The answer to Díaz's question, therefore, is that phylogenetic progression is not comparable to a process of learning in a way that allows anything like to paradox of self-consciousness to get a grip.

PROBLEM 3 It is certainly correct to point out that somatic proprioception and the body-image are susceptible to top-down influence. But it does not follow that the hierarchy of types of self-consciousness is compromised. It is important to distinguish different and dissociable cognitive abilities that are often lumped together under the label "body-image" and treated as if they added up to a single unitary phenomenon. The following types of somatic proprioception are all distinct.

      i) Basic capacities for locomotion and maintaining posture.

      ii) Visuo-motor calibration in action.
      iii) Abilities to locate sensations in the body.
      iv) Understanding of bodily parts and functions.
      v) Knowledge of bodily structure.
      vi) Affective responses to the body.
      vii) Maintenance of homeostasis

The examples of top-down influence that spring most easily to mind fall under types (iv). (v) and (vi). As I tried to bring out in Chapter 6 of my book, however, it is to (i), (ii) and (iii) that we must look for the clearest examples of nonconceptual self-consciousness - and these are not areas in which top-down influence seems widespread.



BERMUDEZ, J. L. 1998. The Paradox of Self-Consciousness. Cambridge MA. MIT Press.

BERMUDEZ, J. L. and ELTON, M. E. (Eds.) 2000. Special issue of Philosophical Explorations (January 2000) on the distinction between personal and sub-personal levels of explanation.

GALLISTEL, C: R: 1990. The Organization of Learning. Cambridge MA. MIT Press.

top | back to symposium index | Díaz's paper