The beginning Socratic injunction "Know thyself" dates to the very beginning of the Western intellectual tradition, and self-knowledge in its many forms has been a central concern ever since. A wide variety of cognitive states fall under its conceptual umbrella. The sort of self-knowledge acquired through cognitive psychotherapy, for example, might include explicit beliefs about the motives of one's behavior couched in the concepts of folk psychological intentional discourse and available for application in the control or modification of future personal action. Alternatively, the Chomskyan self-knowledge that humans are alleged to have of the structure of their language processing systems is largely implicit, subpersonal, innate, and limited in application to the specific process of language-learning.

The types of self-knowledge vary along four main parameters:

-- Content
-- Manner or mode of representation
-- Domain of application
-- Means of acquisition

Although the four are importantly interdependent, each presents its own set of issues.

The content of self-knowledge essentially concerns what a cognitive agent knows about its own nature or organization. Though paradigm cases involve knowledge about one's psychological nature, any property or feature of the agent can be its object. One can have self-knowledge of the size, state, and orientation of one's body as well as of one's beliefs and emotions. Despite this breadth, it is the cases of metapsychological cognition that are of greatest interest; for it is they that provide the opportunity for radical increases in mental sophistication. It is virtually impossible to create a system with highly sophisticated mental abilities without building in a significant degree of metapsychological understanding. This point can be overlooked if one focuses exclusively on cases of explicit human self-knowledge of the sort that could be verbally reported. However, if one recognizes the wealth of implicit self-knowledge that must be implicitly embodied in the systems of self-monitoring and self-regulation required by any sophisticated cognitive agent, it becomes readily apparent that self-knowledge of one sort or another will be the pervasive and central feature of all but the most minimally minded systems. Even simple organisms and robots require some measure of self-knowledge to function as cognitive actors. Without knowledge of one's goals and abilities, it would be impossible to carry out multiple stage actions. And no learning would be possible in a system that totally lacked understanding of the function and organization of the processes to be modified. The requisite self-knowledge might all be implicitly embedded in the structure of the learning mechanisms, but that organization in itself counts as a form of self-knowledge insofar as it adaptively reflects the nature of the processes that are changed through learning.

Explicit human self-knowledge nonetheless remains of great interest to philosophers and psychologists. Adult humans have unique abilities to reason about their mental lives and generally reliable insight into their inferential processes, motives, and preferences. This enhances their ability to regulate their mental lives and to interact with others in social contexts. Empirical studies have nonetheless shown glaring gaps in human metacognitive powers; in specific test situations adult subjects show ignorance of the factors governing their choices, their sources of information, and the rules underlying their reasoning.

The degree to which explicit self-knowledge may be present in children or in nonhuman primates is controversial. Children under the age of three fail to distinguish between their own beliefs about a situation and those that an agent with different access would have; they seem unable to conceptualize their own view of reality as such, that is, as one among many possible views. Tests, especially those involving deceptions, on chimpanzees seem to show some grasp of explicit mental concepts, but only of the most limited sort. Other studies, such as those using mirror recognition tasks, give evidence that chimps and orangs have some type of self concept, but again of only a limited sort. Further research is needed to resolve these issues. In a surprising twist, the psychologist Alan Leslie has proposed that an inability to engage in METACOGNITION is the primary deficit in AUTISM and the source of most of the disabling symptoms associated with it.

Cases of self-knowledge vary not only in their objects but also in how they conceptualize or categorize their objects. Explicit human self-knowledge of the aims of one's behavior is likely to be conceptualized in the folk psychological notions of belief, desire, and intention, but such concepts are not likely to figure in the self-knowledge one's retrieval system has of the structure and organization of one's memory nor in a cat's knowledge of its current needs and goals. Indeed, there may be no public-language words or concepts that adequately capture the content of such implicit knowledge. But insofar as it partitions its pyschological objects in similarity classes and generates aptly matched responses, even such procedural self-knowledge must involve categorization or conceptualization.

Self-knowledge can be explicitly represented and stored as is likely the case with the propositional knowledge that humans have of their intentions and beliefs. The psychologist Philip Johnson-Laird has argued that the planning and control of intentional action requires a self model that explicitly represents one's goals, abilities, options, and current state. But significant self-knowledge can also be implicitly represented in the structure of a network or in the organization of a metacontrol system as is probably true of that embodied in many learning processes. The mechanism by which rats learn to avoid foods that have been followed by bouts of nausea several hours after feeding provides an apt example. One need not suppose that the rat has any explicit awareness of the processes that regulate its feeding behavior or taste sensations, yet its nervous system clearly carries information about those factors as shown by its ability to alter those mechanisms in just the way needed to produce the desired behavioral change.

Procedurally embodied self-knowledge will often be more limited in its scope of application; the metapsychological understanding carried in a learning process may have no impact outside the context of the specific modifications it is designed to produce. The rat can not reflect on the organization of its feeding mechanisms nor make open-ended use of that information, as one might if one had explicit propositional knowledge of them. The difference, however, is one of degree. Procedural self-knowledge can have a relatively broad range of application, and even explicit beliefs about one's mental state are limited in their impact to some degree by the larger context within which they occur.

Self-knowledge can arise in many ways. Traditional Cartesian mentalism treated the mind as fully transparent and open in all its significant properties to a faculty of conscious INTROSPECTION or reflection, which was conceived of by later empiricists as a form of inner perception. Though introspection is now regarded as fallible, incomplete, and theory-laden, it nonetheless remains a major source of self-knowledge. Though internal monitoring, like external monitoring through the senses, is subject to error, it still provides a regular ongoing supply of information about the current state and operation of at least some aspects of one's mind.

The development of self-knowledge depends on both cultural and biological sources. Some theorists argue that the child's mastery of mental concepts involves building a behaviorally based theory of mind; others see it more as a matter of projection from first person-based concepts. In either case, both innate information and culturally based learning will be involved. Children must acquire from their social context many of the concepts needed to categorize their mental states and processes, but folk psychology is likely also to embody an innate scheme of mental categories. Moreover, the neural mechanisms that underlie introspection probably depend on innate implicit self-knowledge in the same way that our perceptual processes depend on such knowledge about the environment.

Though self-knowledge has many obvious benefits, it may not always be adaptive. Ignorance of one's limitations may enhance one's ability to mobilize oneself to action, and a lack of self-knowledge may signal a happy freedom from narcissistic self-absorption. This creates a new puzzle about self-deception. The old puzzle was to explain how deceiving oneself was even possible, given the apparently self-defeating identity of deceiver and deceived. If self-deception is often adaptive, a new more normative problem arises. Must we abandon our intuition that self-deception in itself is a bad thing? Must we qualify the Socratic "Know thyself" with the proviso "but only when it's useful"?

See also

Additional links

-- Robert van Gulick


Beckoff, M., and D. Jameison. (1996). Readings in Animal Cognition. Cambridge, MA: MIT Press.

Bennett, J. (1988). Thoughtful brutes. Proceedings and Addresses of the American Philosophical Association 62:196-211.

Chomsky, N. (1980). Reflections on Language. New York: Pantheon.

Garcia, J. A., and R. A. Koelling. (1967). A comparison of aversions caused by x-rays, toxins and drugs in rats. Radiation Research Supplement 7:439-450.

Goldman, A. (1993). The psychology of folk psychology. The Behavioral and Brain Sciences 16:15-28.

Gopnik, A. (1993). How we know our minds: The illusion of first-person knowledge of intentional states. Behavioral and Brain Sciences 16:15-28.

Johnson, M., S. Hashtroudi, and D. Lindsay. (1993). Source monitoring. Psychological Bulletin 114:3-28.

Johnson-Laird, P. (1983). Mental Models. Cambridge, MA: Harvard University Press.

Lyons, W. (1986). The Disappearance of Introspection. Cambridge, MA: MIT Press.

McLaughlin, B., and A. Rorty. (1988). Perspectives on Self-Deception. Berkeley: University of California Press.

Nisbet, R., and T. Wilson. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review 84:231-259.

Rorty, A. (1975). Adaptivity and self-knowledge. Inquiry 18:1-22.

Wellman, H. (1990). The Child's Theory of Mind. Cambridge, MA: MIT Press .