Justification (from MITECS)

Philosophers distinguish between justified and unjustified beliefs. The former are beliefs a cognizer is entitled to hold by virtue of his or her evidence or cognitive operations. The latter are beliefs he or she is unwarranted in holding, for example, beliefs based on sheer fantasy, popular superstition, or sloppy thinking. Some justified beliefs are based on scientific findings, but scientific beliefs do not exhaust the class of justified beliefs. Ordinary perceptual and memorial beliefs, such as “There is a telephone before me” and “Nixon resigned from the Presidency,” are also normally justified.

A belief’s justification is usually assumed to be something that makes it probable that the belief is true. But justified beliefs are not guaranteed to be true. Even sound scientific evidence can support a hypothesis that is actually false. Almost all contemporary epistemologists accept this form of fallibilism.

There are three principal approaches to the theory of justification: foundationalism, coherentism, and reliabilism. The historical inspiration for foundationalism was RENÉ DESCARTES (1637), who launched the project of erecting his system of beliefs on solid, indeed indubitable, foundations. Most contemporary foundationalists reject Descartes’s insistence on indubitable or infallible foundations, because they doubt that there are enough infallible propositions (if any) to support the rest of our beliefs. On their view, the core of foundationalism is the notion that justification has a vertical structure. Some beliefs are directly or immediately justified independently of inference from other beliefs, for example, by virtue of current perceptual experience. These beliefs are called basic beliefs, and they comprise the foundations of a person’s justificational structure. Nonbasic justified beliefs derive their justification via reasoning from basic beliefs. It has been widely rumored that foundationalism is dead, perhaps because few people still believe in infallible foundations. But most epistemologists regard weak versions of foundationalism—those that acknowledge fallibility— as still viable.

Coherentism rejects the entire idea of basic beliefs, and the image of vertical support that begins at the foundational level. Instead, beliefs coexist on the same level and provide mutual support for one another. Each member of a belief system can be justified by meshing, or cohering, with the remaining members of that system. Beliefs get to be justified not by their relationship with a small number of basic beliefs, but by their fit with the cognizer’s total corpus of beliefs.

Reliabilism, a theory of more recent vintage, holds (in its simplest form) that a belief is justified if it is produced by a sequence of reliable psychological processes (Goldman 1979). Reliable processes are ones that usually output true beliefs, or at least output true beliefs when taking true beliefs as inputs. Perceptual processes are reliable if they typically output accurate beliefs about the environment, and reasoning processes are (conditionally) reliable if they output beliefs in true conclusions when applied to beliefs in true premises. Some types of belief-forming processes are unreliable, and they are responsible for unjustified beliefs. This might include beliefs formed by certain kinds of biases or illegitimate inferential methods.

Because reliabilism’s account of justification explicitly invokes psychological processes, its connection to cognitive science is fairly straightforward. To determine exactly which of our beliefs are justified or unjustified, we should determine which of the belief-forming practices that generate them are reliable or unreliable, and that is a task for cognitive science (Goldman 1986). There is no special branch of cognitive science that addresses this topic, but it is sprinkled across a range of cognitive scientific projects.

Although most perceptual beliefs are presumably justified, some applications of our visual object-recognition system are fairly unreliable. Biederman’s (1987) account of visual object recognition posits a process of matching visually detected components of a stimulus with the component types associated with object categories, such as “cup,” “elephant,” or “airplane.” Object recognition is sometimes produced by partial matching, however, as when an occluded or degraded stimulus reveals only a few of its contours. When partial matching is unreliable, beliefs so formed are unjustified. An example of unreliable reasoning is the overuse of confirming information and the neglect of disconfirming information in drawing covariational conclusions: “I am convinced you can cure cancer with positive thinking because I know somebody who whipped the Big C after practicing mental imagery.” People focus on instances that confirm their hypothesis without attending to evidence that might disconfirm it (Crocker 1982; Gilovich 1991). The search for evidence can also be biased by desire or preference. If we prefer to believe that a political assassination was a conspiracy, this may slant our evidence collection. In one study, subjects were led to believe that either introversion or extroversion was related to academic success. Those who were led to believe that introversion was predictive of success (a preferred outcome) came to think of themselves as more introverted than those who were led to believe that extroversion was associated with success (Kunda and Sanitioso 1989).

Proponents of foundationalism and coherentism often reject the relevance of experimental science to the theory of justification (e.g., Chisholm 1989). But even these types of theories might benefit from psychological research. What makes a memory belief about a past event justified or unjustified, according to foundationalism? It is the conscious memory traces present at the time of belief. Exactly what kinds of memory traces are available, however, and what kinds of clues do they contain about the veridicality of an apparent memory? This question can be illuminated by psychology. Johnson and Raye (1981) suggest that memory traces can be rich or poor along a number of dimensions, such as their sensory attributes, the number of spatial and temporal contextual attributes, and so forth. Johnson and Raye suggest that certain of these dimensions are evidence that the trace originated from external sources (obtained through perception) and other of these dimensions are evidence of an internal origin (imagination or thought). A characterization of these dimensions could help epistemologists specify when a memory belief about an allegedly external past event is justified and when it is unjustified (Goldman forthcoming).

Epistemological theories of justification sometimes ignore computational considerations, which are essential from a cognitivist perspective. Coherentism, for example, usually requires of a justified belief that it be logically consistent with the totality of one’s current beliefs. But is this logical relationship computationally feasible? Cherniak (1986) argues that even the apparently simple task of checking for truth-functional consistency would overwhelm the computational resources available to human beings. One way to check for truth-functional consistency is to use the familiar truthtable method. But even a supermachine that could check a line in a truth-table in the time it takes a light ray to traverse the diameter of a proton would require 20 billion years to check a belief system containing only 138 logically independent propositions. Thus, the coherence theory implicitly requires a humanly impossible feat, thereby rendering justification humanly unattainable (Kornblith 1989). Attention to computational feasibility, obviously, should be an important constraint on theories of justification.


—Alvin I. Goldman


Biederman, I. (1987). Recognition-by-components: A theory of human image understanding. Psychological Review 94: 115– 147.

Cherniak, C. (1986). Minimal Rationality. Cambridge, MA: MIT Press.

Chisholm, R. (1989). Theory of Knowledge. 3rd ed. Englewood Cliffs, NJ: Prentice Hall.

Crocker, J. (1982). Biased questions in judgment of covariation studies. Personality and Social Psychology Bulletin 8: 214–220.

Descartes, R. (1637). Discourse on Method.

Gilovich, T. (1991). How We Know What Isn’t So. New York: Free Press.

Goldman, A. (1979). What is justified belief? In G. Pappas, Ed., Justification and Knowledge. Dordrecht: Reidel. Reprinted in Goldman (1992).

Goldman, A. (1986). Epistemology and Cognition. Cambridge, MA: Harvard University Press.

Goldman, A. (Forthcoming). Internalism exposed. In M. Steup, Ed., Knowledge, Truth, and Obligation: Essays on Epistemic Responsibility and the Ethics of Belief. New York: Oxford University Press.

Johnson, M., and C. Raye (1981). Reality monitoring. Psychological Review 88: 67–85.

Kornblith, H. (1989). The unattainability of coherence. In J. Bender, Ed., The Current State of the Coherence Theory. Dordrecht: Kluwer.

Kunda, Z., and R. Sanitioso. (1989). Motivated changes in the self- concept. Journal of Experimental Social Psychology 25: 272–285.

Further Readings

Alston, W. (1989). Epistemic Justification. Ithaca, NY: Cornell University Press.

Bonjour, L. (1985). The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press.

Goldman, A. (1992). Liaisons: Philosophy Meets the Cognitive and Social Sciences. Cambridge, MA: MIT Press.

Harman, G. (1986). Change in View. Cambridge, MA: MIT Press.

Lehrer, K. (1990). Theory of Knowledge. Boulder, CO: Westview Press.

Pollock, J. (1986). Contemporary Theories of Knowledge. Totowa, NJ: Rowman and Littlefield.

Steup, M. (1996). An Introduction to Contemporary Epistemology. Upper Saddle River, NJ: Prentice Hall.

Last edited on Wednesday, September 3, 2003 1:14:49 pm.