BACHELORS, BUCKYBALLS, AND GANDERS:
SEEKING ANALOGUES FOR DEFINITIONS
OF
“CRITICAL THINKER”
Stephen P. Norris
Memorial University of Newfoundland
Several philosophers of education have offered definitions of
“critical thinker.” In this paper1 I shall examine the
definitions of three philosophers, Robert Ennis, John McPeck,
and Harvey Siegel, with a view to evaluating the significance
of those definitions in the promotion of critical thinking. My
conclusion is that for definitions of “critical thinker”
to be significant they must be capable of serving as the basis
of empirical research on the nature and fostering of critical
thinkers.
Before I look at the definitions at all I shall develop, in the first part of the paper, distinctions among types of nouns that help highlight the different stances taken by philosophers of education who offer definitions of “critical thinker.” The work of Schwartz2 on nominal and natural kind terms will guide this first section. In the second section, I shall explore the implications of taking “critical thinker” to be either a nominal or natural kind term. The third section examines how the three named philosophers of education regard their definitions of “critical thinker.” Finally, I shall argue for what I consider to be the best way to take definitions of “critical thinker,” and suggest what implications adopting my view would have for philosophers of education working in the area.
It has been advocated for at least three centuries, from Locke to Lewis to Copi, that the meaning of common nouns, exemplified by “bachelor,” “sister,” “table,” “triangle,” and “key,” is explained well by postulating that the nouns have associated with them an intension and an extension. The intension of a term, sometimes called the concept associated with it, contains a list of properties that determines the referents of the term. The set of referents makes up the extension. The properties listed in the intension of “bachelor,” for example, might include being male, being unmarried, and being of marriageable age. Having each of the properties in the list is severally necessary and jointly sufficient for being an entity in the extension of the term. Thus, intension determines extension; if Frank has all the properties listed in the intension of “bachelor,” then he is a bachelor.
Since the extension of a term is the set of entities that have the properties listed in the intension, the extension of a term is determined empirically through a procedure of matching the properties listed to entities in the world. The properties that constitute the intension of a term depend upon the conventional meaning of the term. Thus, the intension also is determined empirically by analyzing language use. The intension is semantically associated with the term. If a community of language users alters the properties listed in the intension of a term, then it has altered the term’s meaning, as appears to be happening with the meaning of “bachelor.”
One claimed attraction of this theory is that it explains how words acquire a purchase on the world: they do it via the reference relation, which is determined empirically by matching lists of properties to entities. Another touted attraction is that the theory allows us to distinguish between analytic and empirical truths associated with terms. “Bachelors are unmarried” is true analytically, that is, by definition, because “being unmarried” is a property listed in the intension of “bachelor.” “Bachelors earn greater than average-sized incomes” is, if true, true empirically, that is, according to whether or not it corresponds to what occurs in the world, because “earning greater than average-sized income” is not part of the intension of “bachelor.”
Because of its simplicity, explanatory power, and promise to clarify the grounds for the truth of various types of statements, this theory of the meaning of common nouns has received wide acceptance. I shall call the nouns for which the theory does seem to work well, “nominal kind terms.” Schwartz proposes that terms for artifacts (“chair,” “book,” “hammer”), distinctions of rank (“president,” “corporal,” “assistant professor”), relations between people (“sister,” “nephew,” “sibling”), and legal and ceremonial use (“contract,” “ordination,” “will”) are examples of nominal kind terms.3
Unfortunately, the acceptance has been too wide, because it has been applied to the meaning of all nouns but does not work well for all. Consider the proper name “Shakespeare.” Paralleling the analysis of “bachelor,” it is sometimes claimed that there is an intension and extension (in this case an extension containing one entity) associated with this term, and that the intension determines the extension. For example, we might associate with Shakespeare the property that he wrote the play Hamlet.
However, the parallel with common nouns breaks down when we consider the statement, “Shakespeare is the author of Hamlet,” and ask whether the statement is analytically or empirically true. If the theory outlined above for common nouns is the correct theory of the meaning of “Shakespeare,” then the statement must be analytically true, since being the author of Hamlet would be part of the concept associated with the term.
However, a theory of meaning that yields this result must be an incorrect theory of the meaning of proper names. We can well imagine, for instance, learning some day that Shakespeare (that very same person to whom we now refer) did not write Hamlet at all. We may learn that Francis Bacon wrote the play. Since this is a conceivable outcome, unlike the outcome of finding a married bachelor, being the author of Hamlet, if it is associated with the proper name “Shakespeare,” must be contingently associated with it. Furthermore, finding out that Bacon did write Hamlet would not change necessarily the referent of “Shakespeare.” The name might still refer to that English poet and dramatist who lived between 1564 and 1616.
Are these latter properties, then, part of the intension of “Shakespeare”? Again, they are not, since they are, if true, empirical truths, not truths by definition. We can imagine discovering that Shakespeare was born or died on different dates, or that he was not English but French, and moved to England from France as a child. Upon analysis, it can be seen that proper names do not have meanings in the sense of common nouns. That is, they do not have concepts associated with them; there are no analytic statements associated semantically with proper names that determine the entities that fall into their extensions. Proper names have associated with them, instead, a set of identifying descriptions that are empirically defeasible. We might say that these descriptions form a conception, rather than concept, of the entity named. The conception is used to help pick out the referent, but not to determine it. If, however, in a given case, enough of the identifying descriptions associated with a proper name are found to be false, then the referent of the name may be called into question. It is conceivable, for instance, that an historical analysis of the use of the name “Shakespeare” result in the conclusion that “Shakespeare” refers to Bacon.
There is also a set of common nouns, exemplified by “tiger,” “gold,” “water,” and “human,” whose meaning cannot be explained by the theory of meaning of nominal kind terms. Rather, these terms behave like proper names. Consider “gold,” and what properties might be in its intension. Plausible candidates are being yellow, malleable, metallic, electrically conductive, and valuable. However, contrary to the case of “bachelor” and the property of being unmarried, none of these properties are associated semantically with the term. It is conceivable that gold could be found or created that has none of these properties: this new gold may not be metallic, but rather some other structural form; we may someday be able to produce forms of gold that are not malleable or not electrically conductive; it is already known that, when finely divided, gold may be black, ruby, or purple. Therefore, yellowness, malleability, conductivity, and so on, are identifying descriptions of gold, as being the author of Hamlet is an identifying description of Shakespeare, and not part of the concept of gold. For something to be gold, it is not the case that it must, by definition, have these properties. The descriptions are used to pick out gold, but they do not define what it is. Hence, “gold is malleable” is not analytically true. If true, it is an empirical truth.
“Carbon” is a more interesting example. Carbon is very soft or very hard, black or colorless, opaque or transparent, electrically conductive or electrically insulative, depending upon its allotropic form — graphite or diamond. Having a set of disjunctive properties such as this does not fit the traditional theory of the meaning of nominal kind terms. The case of carbon shows that, for a particular class of entities, phenomenal properties can be misleading indicators of their nature. If “carbon is very soft” were analytically true, then it would be impossible to discover carbon that is very hard, just as it is impossible to discover a married bachelor. It would also have been impossible to have discovered recently (as has been done) a third allotropic form of carbon — buckminsterfullerene or buckyball — that has yet another set of phenomenal properties.
Rather than take “gold” or “carbon” to be terms with associated concepts that pick out their extensions, we might take them to be singular referring terms, as we take the proper name “Shakespeare.” But, to what do they refer? One plausible answer is that they refer to an underlying trait4, which it is the job of empirical research to identify. The current view is that gold is a substance composed of atoms of atomic number 79, and that such atoms are characterized by having particular numbers of electrons, protons, and neutrons. If gold, meaning the stuff to which the word “gold” refers, has this underlying trait (I say “if,” since the trait is defeasible), it is a trait that it must have out of empirical necessity, because with any other structure it would not be gold, but some other substance. Again, if gold has this trait, “gold” refers to it, and the extension of “gold” consists of those entities, and only those entities, that have the trait. This is why instances of graphite, diamond, and buckminsterfullerene all fall into the extension of “carbon”: they all have the same underlying trait — atoms of atomic number 6 — even though they have contradictory phenomenal properties. If current beliefs change about the phenomenal properties of gold and carbon or about the traits that underlie gold and carbon, the extensions of “gold” and “carbon” need not change, just as the extension of “Shakespeare” need not change if our beliefs about Shakespeare change. For instance, we may give up our belief in atoms, electrons, protons, and neutrons, and thus not say any more that carbon is that substance composed of atoms of atomic number 6, yet “carbon” may still refer to the same stuff it does now. I shall call common nouns whose extensions are determined by the presence or absence of underlying traits strict natural kind terms.
In addition to nouns that derive their extensions via semantically related properties and empirically related underlying traits, there are those that derive their extensions via both routes. An example is “gander.” Maleness is semantically associated with “gander,” so the statement “ganders are male” is analytically true. However, that ganders have webbed feet, bills, feathers, and a certain voice are identifying descriptions of ganders, and thus are empirically, but not semantically, related to “gander.” We could imagine ganders that do not have webbed feet, or any feet at all for that matter, do not have bills, do not have feathers, and have no voice. These imagined ganders may, for instance, be the offspring of geese that are exposed to radioactive fallout. However, once we identify the underlying species trait that makes ganders, maybe it is something about the structure of their DNA, then ganders have that trait out of empirical necessity. Common nouns whose extensions are determined by both semantically related properties and empirically related underlying traits shall be called non-strict natural kind terms.
Into which of the above categories might “critical thinker” be placed? Churchland5 has argued that “thinker” is “quite possibly, bordering on probably,” a strict natural kind term. He postulates further that thinkers are the same natural kind as are living things, and that the underlying trait of this kind is defined by the physics of non-equilibrium thermodynamics: “a living thing is a dissipative system, a semi-closed local entropic minimum, whose internal negative entropy filters out further negative entropy from the energy flowing through it.”6 While the precise meaning of this trait description may not be grasped by most of us (it basically says that living things maintain a structural order in a universe that tends to favor chaos; that life runs counter to the flow of time), the fact that someone has suggested an underlying trait for the entities that fall into the extension of “thinker” is what is important here.
Churchland, at least, should think it meaningful to ask whether the properties that are commonly associated with critical thinkers (open-mindedness, disposition to seek reasons, ability to judge credibility, and so on) are properties semantically related to a nominal kind term, identifying descriptions empirically related to a natural kind term, or underlying traits possessed by the entities to which a natural kind term refers. Maybe some of the properties are semantically related to “critical thinker,” others empirically related identifying descriptions, and maybe somewhere there is an underlying trait. Does “critical thinker” function in educational theorizing mostly like “bachelor,” mostly like “buckyball,” or mostly like “gander”? A possibly more important question concerns how it ought to function in our educational theorizing. These are non-trivial and theoretically important questions.
If “critical thinker” is a nominal kind term, then it makes sense for theorists to differ over its meaning according to their individual programmatic agendas. Under this scheme of things, the properties associated with “critical thinker” would be associated semantically with it; changing the list of properties would amount to changing the concept related to the term, and thus the class of individuals that comprises its extension; and deciding on the list of properties would be a matter of how theorists wanted language to be used in order to serve their educational agendas. This decision would be influenced primarily by the theorists’ values. Depending upon their values, theorists might make different recommendations about whom should be classified as critical thinkers.
If “critical thinker” is a strict natural kind term, assigning individuals to the class could not be solely a value issue. The extension of “critical thinker” would not be determined by a set of negotiable, semantically related properties. The extension would include only those individuals who had the trait that underlies critical thinkers. This trait would be non-arbitrary, and the task of empirical educational research to discover. The empirical research might begin from an ostension or description that fixes the referent of “critical thinker.” Reference fixing goes more or less like this:
A decision is taken that critical thinkers are the kind instantiated by almost all of such-and-such sample of individuals. From this point, empirical research would explore the underlying traits of the individuals in the sample. The possible outcomes of such research are several. On one possibility, a trait that underlies all the individuals is found. On another possibility, a trait is found to underlie a portion of the sample, and another trait to underlie another portion. A third possibility is that the individuals are found to have nothing that could be called a common underlying trait.As the final option, it is possible that “thinker” is a strict natural kind, as Churchland has postulated, but that “critical thinker” refers to a negotiated division within that kind, much like “gander” refers to a negotiated division of geese. If “critical thinker” is a non-strict natural kind term of this sort, then there will be properties associated with “critical thinker” that are negotiable according to value orientations, but an underlying trait that is the task of empirical educational research to discover.
Ennis provides a list of abilities and dispositions that, he proposes, critical thinkers have.7 Does saying that critical thinkers have these abilities and dispositions express a series of analytic truths according to Ennis? If so, and if for Ennis this is all there is to the meaning of “critical thinker,” then “critical thinker” would be a nominal kind term for Ennis. On the other hand, does his list provide a set of identifying descriptions of critical thinkers that, if true, are empirical truths? If so, and if Ennis believes these are not properties semantically related to “critical thinker,” then “critical thinker” would be a strict natural kind term for Ennis. Finally, does the list or some items on the list entail analytic truths about critical thinkers, while allowing that critical thinkers have some underlying trait that is their nature? If so, then “critical thinker” would be a non-strict natural kind term for Ennis.
I find it difficult to answer these questions about Ennis’s ideas. I am inclined to think he believes that “critical thinker” is a term over which we should try to reach agreement on conventional meaning. Ennis does not believe that he has provided a set of severally necessary and jointly sufficient conditions for being a critical thinker, but he intends the list to be comprehensive.8 Aiming for a complete list does not suit the task of choosing identifying descriptions for a natural kind term, because all that is needed is a set sufficiently discriminating to home in on the desired referent. A plausible possibility is that Ennis takes “critical thinker” to be a nonstrict natural kind term, and that his theorizing has been an attempt to influence the meaning of the conventional part, namely, “critical.” One reason for trying to influence what people take the conventional part to mean would be a belief that the identifying descriptions of critical thinkers can be a productive focus for educational interventions.
What about Siegel; how would he classify critical thinkers, whom he sees as persons “appropriately moved by reasons.”9 Unlike Ennis, Siegel does not provide long lists of abilities and dispositions held by critical thinkers. I believe that Siegel takes critical thinkers to be a strict natural kind. He claims that “a critical thinker is…a certain sort of person…. Just as sugar has the disposition to dissolve in water while still in the sugar bowl, so does the critical thinker have the dispositions, habits of mind and character traits we have considered while not engaged in reason assessment.”10 It seems fairly clear from this statement that Siegel believes that critical thinkers have underlying traits characteristic of them, and that only critical thinkers have those traits. Whether or not Siegel also believes that there are characteristics that are semantically related to “critical thinker” is unclear to me at this time. If he does, I would modify my assessment to conclude that he takes critical thinkers to be a non-strict natural kind.
How about McPeck? It seems clear that he takes “critical thinker” to be a nominal kind term. McPeck11 charges Norris,12 writing in the following passage about “reasoning” as a natural kind term, with confusing philosophical and scientific questions:
The first question [whether “reasoning” denotes a particular process, performance, or type of achievement] would concern the denotation of “reasoning,” which would involve the same sorts of exploration used to determine the denotation of any natural kind term. It would be necessary to carry out scientific investigations into the underlying nature of reasoning.13While Norris clearly states his position that “reasoning” (and by implication “critical thinking,” since that was the context of the discussion) is a natural kind term, McPeck just as clearly states his position: “the denotation, qua denotation, of common terms like “reasoning” is a conceptual question par excellence, and has nothing whatsoever to do with scientific investigation.”14
It is not a matter of taste with no consequences for educational practice which position on the definition of “critical thinker” is adopted. One purported aim of critical thinking theorizing is to make school students better critical thinkers, that is, to effect change in the world. In this regard, we can expect some practices to work and others not to work, but, it seems to me, those that have the poorest chance of success are those that are based on inadequate theories of human learning. A minimal condition of adequacy for a theory of human learning is that its central terms refer. For instance, the adequacy of a theory of learning that used as a central term, “innate intelligence,” would be reduced if innate intelligence does not exist. A central term of many critical thinking learning theories is “thinking disposition.” The adequacy of these theories is reduced if thinking dispositions do not exist.
The strict nominalist stance, with its anti-empirical approach to determining denotation, seems to be the approach to defining “critical thinker” that is least likely to meet this minimal condition of a critical thinking learning theory. Virtually all major theories of meaning assume that empirical research is necessary for determining denotation. In 1892, Frege15 showed us that conceptualization alone would never have led to the truth that “the morning star” and “the evening star” denote the same entity. In 1905, Russell16 showed that determining the truth value of “The King of France is bald,” depends upon determining whether there is an entity that possesses the property of being the King of France. In 1966, Donnellan17 showed that the meaning of definite descriptions depends upon empirically checking whether the descriptions are attributing properties to entities or merely referring to entities. Thus, if we want the term “critical thinker” to denote, which is necessary if we want any theory that employs the term as a central concept to have anything to do with educating people, then its definition cannot be derived using solely conceptual analysis.
The demand for denotation by the central terms in a theory of critical thinking learning, in particular the term “critical thinker,” seems to point us, then, towards natural kind terms. But how is the denotation of such terms determined? The simple answer is by doing empirical research. I cannot hope to outline a full description of the nature of that research in this paper. However, if we consider for the moment the denotation of “critical thinker,” it would be a worthwhile approach first to study individuals who seem to exemplify most clearly and centrally our pre-scientific notion of critical thinkers. Such individuals might include some of the best artists, literary critics, scientists, medical doctors, engineers, nurses, politicians, and so on. The initial focus of the research might be on what, if anything, is common about the thinking of these individuals when they are engaged in thinking in their fields.
I cannot imagine how this research might turn out, and I do not have the space here to propose methods for conducting it. However, current theories of critical thinkers suggest some questions that might be explored. For instance, Ennis’s theory suggests asking whether the individuals identified as critical thinkers use in their thinking the principles of thought that are central to his theory. Siegel’s theory suggests asking whether the individuals identified as critical thinkers are disposed to, for instance, seek evidence for their beliefs more than individuals not identified as critical thinkers.
Thus, philosophers’ theorizing can be at the basis of empirical critical thinking research: it can provide hypotheses that empirical researchers test; it can provide explanations of findings that empirical work yields. In order to play this role, however, philosophers of education must frame their theories of critical thinking so that the empirical implications of them are made clear, or must work with empirical researchers in deriving such implications. This means taking seriously the sorts of evidence that would be relevant to testing their theories. For instance, if a theory of critical thinking includes a postulation of a disposition such as open-mindedness, then we need a clear account of what evidence would count for the presence or absence of that disposition. Such accounts are not available and, in part for this reason I believe, there has been virtually no empirical research on critical thinking dispositions. If a theory of critical thinking includes a postulate that critical thinking is generalizable from field to field, then specifications are needed of the type of evidence that would test the truth of this postulate. Ennis has begun to offer such specifications,18 and, in part as a consequence of Ennis’s efforts, we are beginning to see some empirical work on the generalizability issue.
At the same time, since “critical thinker” is a term that is intended for the prescription of educational goals and practices, it cannot be treated as a strict natural kind term. Philosophers must have some conceptual leeway to build into the concept of critical thinker features that are valuable to education. For instance, Ennis19 once proposed that thinking critically implies thinking morally, while Martin20 argues that too much critical thinking takes place that violates moral standards. It seems to me that this is not the sort of issue that can be settled by empirical research. It depends upon how we want the concept of critical thinker to operate in our educational prescriptions. Thus, my current thinking is that “critical thinker” is a non-strict natural kind term — one that is constrained by the psychology of human beings and also by educational values.
If “critical thinker” is a non-strict natural kind term, then the framers of critical thinking theories should play an active role in specifying what would count as evidence for or against those theories. Since many of the theorists of critical thinking are philosophers of education, the empirical scrutiny of such questions as what “critical thinker” denotes may be unappealing. I submit, however, that only by becoming more closely involved in empirical research will philosophers of education serve the educational goal of fostering critical thinking development. On the other hand, I hope that such an approach by philosophers of education will be welcomed by empirical researchers, because only if their work is founded on sound theory will it foster critical thinking development.
For responses to
this essay, see McPeck, Siegel and Ennis.
1 This paper is based on my chapter, “The Generalizability Question,” in The Generalizability of Critical Thinking, ed. Stephen P. Norris (New York: Teachers College Press, 1992), 1-15. I thank Ki Su Kim, Walter Okshevsky, and Linda Phillips for comments on a previous draft.
2 Stephen P. Schwartz, “Natural Kind Terms,” Cognition 7 (1979): 301-315.
3 Ibid., 311.
4 William K. Goosens, “Underlying Trait Terms,” in Naming, Necessity, and Natural Kinds, ed. Stephen P. Schwartz (Ithaca, New York: Cornell University Press, 1977), 133-154.
5 Paul M. Churchland, “Is Thinker a Natural Kind?,” Dialogue 21 (1982): 236.
6 Ibid., 233.
7 Robert H. Ennis, “A Taxonomy of Critical Thinking Dispositions and Abilities,” in Teaching Thinking Skills: Theory and Practice, ed. Joan Boykoff Baron and Robert J. Sternberg (New York: W.H. Freeman & Co., 1987), 9-26.
8 Ibid., 25.
9 Harvey Siegal, Educating Reason (New York: Routledge, 1988), 32.
10 Ibid., 41.
11 John E. McPeck, Teaching Critical Thinking (New York: Routledge, 1990).
12 Stephen P. Norris, “Thinking about Critical Thinking: Philosophers Can’t Go It Alone,” in Teaching Critical Thinking, ed. John E. McPeck (New York: Routledge, 1990), 67-74.
13 Ibid., 71.
14 Ibid., 92.
15 Gottlob Frege, “On Sense and Reference,” in Readings in Semantics, ed. Farhang Zabeeh, E.D. Klemke, and Arthur Jacobson (Urbana, Illinois: University of Illinois Press, 1974), 117-140.
16 Bertrand Russell, “On Denoting,” in Logic and Knowledge, ed. Robert C. Marsh (New York: Capricorn Books, 1956), 41-56.
17 Keith S. Donnellan, “Reference and Definite Descriptions,” The Philosophical Review 75 (1966): 281-304.
18 Robert H. Ennis, “The Degree to which Critical Thinking is Subject Specific: Clarification and Needed Research,” in The Generalizability of Critical Thinking, ed. Stephen P. Norris (New York: Teachers College Press, 1992), 21-37.
19 Robert H. Ennis, “A Conception of Rational Thinking,” in Philosophy of Education 1979, ed. Jerrold R. Coombs (Normal, Illinois: Philosophy of Education Society, 1980), 3-30.
20 Jane Roland Martin, “Critical Thinking for a Humane World,” in The Generalizability of Critical Thinking, ed. Stephen P. Norris (New York: Teachers College Press, 1992), 163-180.