views updated


Most philosophers have been led by the argument from illusion, by the causal argument, or by the introspective analysis advocated in the sense-datum theory to conclude that our immediate awareness in perception is not, as direct, or commonsense, realism claims, of material objects (of distinct, external physical entities perceptible by different persons at once) but of sensa (private, transitory, probably mental existents that may also be called sensations, sense data, ideas, representations, or impressions). Once this position is adopted, a serious difficulty arises concerning the nature and status of material objects. Representative realism claims that they exist external to us and cause the sensa or representations that correspond to them. The notorious difficulty of this view is that if all our direct awareness is concerned with the alleged effects, or sensa, how do we ever find out that material objects exist as their causes or what characteristics they possess? The theory seems to make material objects unobserved, and indeed unintelligible, causes of our perception. Although representative realism, especially in modern versions, tries to deal with this difficulty, it is still widely felt to be unsatisfactory. Therefore, alternative attempts have been made to deal with the problem of the nature of material objects. One such approach, which may loosely be called phenomenalist, is to reduce material objects to sensa, that is, to explain them as consisting solely of sensa or as being primarily groups or patterns of them. This approach results in slightly varying views, and when the term phenomenalism is used, reference is very often intended only to what we here call linguistic phenomenalism.

To introduce these variants of phenomenalism, we may consider one central problem that faces any attempt to reduce material objects to sensa, namely, the fragmentariness of perception. Any material object is believed to exist for long periods when it is not observedfor example, the furniture in an empty room, the beams in the roof, and so onand some objects, such as rocks in Antarctica or under the ocean, may never have been observed. Yet when they are not observed, material objects cause no sensa, have no sensa belonging to them or constituting them. Hence, if material objects are reduced to actual sensa and consist only of them, they must cease to exist when unobserved, and those never observed must never have existed. Worse still, the material objects in a room must apparently come into and go out of existence as one looks at or away from themthe blinking of a human eye can destroy or create them. This seems such an intolerable paradox that George Berkeley, though tempted to say that material objects are simply collections of ideas, had to introduce God as their continuing basis or cause. True phenomenalism, however, can no more allow unobserved divine causes than unobserved material ones.

Problem of Fragmentariness

Several approaches to the problem of fragmentariness may be taken.


One might accept fragmentary existence, though saying it is no insuperable paradox: Objects are no more than groups or patterns of sensa, but owing to the regularity with which the same or similar series of sensa occur, we imaginatively fill in the gaps and falsely suppose that continuously enduring objects exist. This was David Hume's official view. One may say that just as a tune may bridge various pauses when no sound occurs and thus be a pattern of sounds with intervening gaps, so an object may be a group or pattern of sensa and gaps. Nevertheless, the theory is incredible and is only on the fringe of the phenomenalist group of theories. For one thing, it is difficult to see why sensa recur in groups or patterns if nothing exists in between; the existence of some continuant basis or focus of them seems a far simpler and more plausible hypothesis than what would be a series of unexplained coincidences.


Hume himself toyed with the supposition that impressions might exist unobservedthat the gaps might be filled with unsensed sensaand if H. H. Price is right, Hume should have developed this as his official theory. Such a development was explicitly formulated by Bertrand Russell in his Mysticism and Logic, where he gave the name "sensibilia" (singular, sensibile) to these "objects that have the same metaphysical and physical status as sense-data without necessarily being data to any mind." Russell regarded sensibilia as the ultimate constituents of matter; thus, objects consist of systems of sensed sensibilia (that is, sensa) and unsensed ones.

However, he soon abandoned this position, which seems untenable on two main grounds. First, it cannot explain the causal processes in perceiving. How does the sensing of sensibilia bring sensa into being? The evidence of the causal processes and of the conditioning of perception by the state of the nervous system and sense organs suggests that sensa are "generated," that is, brought into being, by events in the brain; this seems incompatible with the view that they existed as sensibilia before they were sensed. Second, what evidence is there of the existence or the nature of sensibilia? One cannot observe that such entities fill gaps between actual sensa; they are just as obscure and hypothetical as the unobserved material objects of representative realism and, in fact, introduce the very difficulty that they were intended to avoid.

factual phenomenalism

Factual phenomenalism attempts to fill the gap between actual sensa with possible ones by defining material objects as groups of actual and possible sensa. This view was originated by J. S. Mill, who held that matter consists of "groups of permanent possibilities of sensation." Unfortunately, this theory also leaves quite obscure what possible sensa could be and adds the further implausibility that the gap-filling entities are purely possibilities and not actualities at all. If taken strictly, this should mean that nothing actually fills the gaps. To say that something, for instance, an accident, is possible implies that it is not actual, though it might be claimed that a possible X is an actual Y ; for instance, the possible winner of a race is an actual horse, in which case once again matter will consist largely of unknown and unobservable entities. The view is also open to many of the objections to phenomenalism stated below.

linguistic phenomenalism

Linguistic phenomenalism sees the basic problem before it in a different light, as one not of stating the constituents of matter but of elucidating the concept of a material object, of defining it in terms of sensa; and it seeks to achieve this not by formal definition but by a "definition in use," that is, by providing translations of statements about material objects into equivalent sets of statements about sensa. Thus, it is intended to show that what is meant by talking about tables, chairs, or similar objects can be expressed solely by talking about sensa; sometimes this is expressed by saying that material objects are logical constructions out of sensa. The underlying position is, in essence, that of Humethat all we know to exist are sensa occurring in various patterns or sequencesbut one main difference lies in the claim that these regular relationships between sensa are not something to be supplemented by imagination but are actually what we indirectly refer to by talking of material objects. Such objects are in fact coordinating concepts, devices that enable us to group and correlate our sense experiences, to identify and to refer to patterns in them.

The other main difference from Hume's position is in the linguistic presentation, the attempt to elucidate the concept by translation into a set of equivalent statements. This is in accordance with the linguistic approach contemporary with the heyday of phenomenalism, and it was held that statements about material objects and statements about sensa are simply two different ways of describing the same set of facts (facts that really concern sense experiences, their patterns and sequences). However, the sets of sensum statements not only are translations but also have a special form. Insofar as the object is observed, they are all categorical, but when it is unobserved, they are hypothetical. Thus, "I see a book on the table" is equivalent to "I have sensa XYZ," where XYZ might stand for "of a rectangular, red, solid-seeming shape on a flat brown expanse." However, "There is a book on the table in the next room" is equivalent to "If you were in the next room, you would have sensa XYZ." This introduces the notion of possibility that was not in Hume and that factual phenomenalism expresses so implausibly. It has the great advantage of expressing the possibility of sensa in the hypothetical form of the statement without suggesting that possible sensa are somehow constituents, perhaps the sole ones, of actual objects. Also of interest is that this approach was anticipated but not developed by Berkeley (Principles, Sees. 3 and 58), and occurs in places in J. S. Mill.

The result is an ingenious theory that transforms the problem of producing a viable alternative to representative realism. If successful, it would be an enormous theoretical economy; it would enable the facts of experience to be accounted for solely in terms of one type of existent, sensa, without any need to go beyond them and postulate other orders of material existence behind them. Indeed, it could further claim to be neutral between the sense-datum and adverbial analyses of sensing, for one could, as Alfred Jules Ayer did, translate material-object statements into statements about "sense contents," a term used to describe how we sense but not to refer to separate entities.

This version of phenomenalism achieved great popularity from about 1930 to 1950, particularly because it was associated with (1) logical positivism and operationalism, the meaning of material-object statements being held to lie in their mode of verification, that is, in the sensum statements that verify them; (2) Russell's analysis of abstract terms, for instance, that space is not an entity but a logical construction out of observations and measurements; (3) a way of dealing with unobserved entities in physics, namely, that electron statements are equivalent to, are logical constructions out of, sets of statements about physicists' observations. However, in the last two cases the data for the construction are prima-facie observations of material objects, and the construction is thus at a different level. Furthermore, the third case gains plausibility from the fact that electrons are agreed to be unobservable; but no such unobservability belongs to tables and chairs.

Difficulties in Phenomenalism

Because of its merits, linguistic phenomenalism became the dominant version of phenomenalism (so much so that the qualification "linguistic" may seem pedantic). All the same, many difficulties soon appeared in it and defied ingenious, almost desperately ingenious, attempts to deal with them. Further, the theory presupposed that our direct awareness is entirely of private sensa; consequently, it has suffered from the recent revival in direct realism. Without questioning that presupposition, we shall now consider the general difficulties in the theory.

lack of equivalence

The original aim of linguistic phenomenalism was to give a fully equivalent translation of a material-object statement into sets of sensum statements, thus proving that it meant no more than is meant by a series of such statements. For various reasons this seems impossible. In the first place, according to the basic supposition of the sense-datum theory that is shared by phenomenalism, there is a different sensum for every different look, sound, feel, or other appearance of a material object. When a dish looks elliptical, one sensum belonging to it is obtained; when it looks round, another one is obtained; when it is felt, yet another; and so on. When one considers all the different points of view from which the dish can be seen and can look different, and then adds all the variations possible for the other senses and for other conditions of lighting and such, it would seem that the number of sensa belonging to the dish, and therefore the number of sensum statements necessary to produce a full analysis or translation of "There is a dish on the table," would be very great. Sometimes it is said that the number would be infinite because the different points of view are infinite in number; but this is dubious, for owing to object constancy, a slight change in point of view would not necessarily mean a different sensum.

At any rate, the list of sensum statements would have to be far longer than can be achieved in practice. Furthermore, if the analysis is really to be adequate, it must be systematic: The sets of sensum statements must be so ordered as to show something of the patterns or correlations that justify the material-object concept; but far from doing this, phenomenalists usually give up after one or two of the sensum statements have been formulated.

Equivalence has also been denied on the ground of difference in form. The original material-object statement is a categorical one, clearly stating that something actually exists. However, the translation is a series of hypothetical statements, and even when the apodoses of these describe experiences, their normal function seems to be either to avoid asserting actual existence (or occurrence) or to convey something quite different, such as a promise or a warning"If you touch that, you will get burned." Indeed, "If you go to the next room, you will see a book on the table" may function as a request or a suggestion that the person go there. Worse still, in the counterfactual statements that form the translation offered about past events, actual existence is denied by implication. Thus, "Pterodactyls lived in the Mesozoic era" would probably be translated "If an observer had been present in the Mesozoic era, he would have had pterodactyl-like sensa." However, there was no observer at that timein fact, no human beings at alland no sensa as we know them. Thus, the assertion of actual existence is replaced in the alleged translation by assertions about what might have happened but did not.

Another bar to the claim of equivalence is that there is not full mutual entailment of original and translation. On the one hand, there might be some illusion or hallucination in which the sensum statements would be true and the material-object statement false: All the red booklike sensa might be present, and yet the object might be a box covered and shaped to look and feel like a book. This can, no doubt, be ruled out in practice by getting enough sensa, especially those resulting from such tests as opening the book, but it is doubtful how far results of such tests are really part of the meaning of the material-object statement and are therefore true features of the translation. On the other hand, the material-object statement might be true and the sensory ones false. There might be a book on the table, and yet you might not get sensa of itthe light might fail, you might be taken ill suddenly or be careless and inattentive, the book might be covered by other objects, and so on. There is a large range of conditions that would have to be stated to ensure the truth of the sensum statement. This is particularly true if the object is a small one: "There is a needle in this haystack." If you looked, would you get the needlelike sensa?

impurity of analysis

A troublesome practical difficulty facing phenomenalists is that it is impossible to specify more than a few sensa without recourse to material-object language (and not always then). Since in considering a book, the formula "sensa of a rectangular, red, solid-seeming shape on a flat brown expanse" would not differentiate the book from, say, a chocolate box, the temptation is to say "a red, rectangular, booklike sensum." But then one no longer has a translation, and the analysis is impure; it is like saying that in French cheval means an animal of a cheval -like nature. Most phenomenalists succumb to this temptation and blame it on the poverty of language, which was designed for speaking about material objects; they say, not very convincingly, that they could invent a proper terminology for describing sensa accurately but that it would take too long.

Another type of impurity in phenomenalistic analyses lies in the protases of the hypotheticals, where reference is normally made to observers and landmarks, for example, "If you go to the next room, you will get sensa XYZ." Even if only your body is a material object, you are at least not a sensum; and similarly, the room is physical and material. Thus, such a hypothetical statement is not a pure sensum statement. Even giving directions by compass points, for example, "If you look north ," would seem to involve some dependence on material objects, such as the sun or a compass. Ayer suggested an ingenious way out of this difficulty: Instead of mentioning the observer and others, you describe the available sensa of the room or location, thus getting "Given sensa ABC, then sensa XYZ are obtainable," where ABC are "interior-of-roomlike sensa" and XYZ are "booklike sensa." (This also slightly mitigates the difficulty about standing conditions mentioned with respect to mutual entailment: If roomlike visual data are given, at least there is light enough to see large objects.) But once again, specifying the roomlike data without mentioning the room, though perhaps theoretically possible, presents great practical difficulties that no one has tried to surmount. Nevertheless, this second impurity problem has at least been reduced to the first one.

publicity and persistence of objects

In view of the great difficulties facing any attempt at a fully equivalent and pure translation, the phenomenalist may modify his aims. He may say that by producing a few sentences of the translation and by using such short cuts as "booklike sensa" he can show the form a full analysis would take; he can give a schema or blueprint of it sufficient to show that a material-object statement really means no more than a set of sensum statements and to reveal the kind of relation between sensa that justifies the material-object concept. Others would argue that this is to abandon the real aim of phenomenalism: Unless one produces a fully equivalent translation, one cannot be sure that there is not some characteristic of material objects that cannot be rendered in terms of sensa. This objection is supported by drawing attention to several features of the ordinary concept of a material object that seem particularly resistant to phenomenalist analysis.

The first of these are the publicity of material objects (the fact that many people can perceive them at once) and their persistence or relative durability. Sensa are private and transitory, so how can statements about them convey the meaning of statements about objects? A phenomenalist answer would be that all we mean or are entitled to mean by saying that an object is perceived by two people at the same time is that they simultaneously sense similar sensa. This can be formulated as: Observer A has sensa XYZ at time t ; observer B has sensa XYZ also at time t ; and both sets of sensa are located similarly with respect to other background sensa. The analysis can be supported by saying that when B senses visual and tactile data describable as data of his touching the object, then A gets visual data describable as data of B touching it. As to the persistence of objects, all this amounts to is that sequences of similar data recur. In development of this point, Hume claimed that it involves constancy (recurrence of the same data each time you look) and coherence (sequences of data changing in an orderly manner); Ayer, however, put most emphasis on the recurrence of reversible series of data, as when you look round the room and then back again.

But these answers are inadequate for the following reasons.

  1. They make the analysis impure by reference to observers: The whole point in the publicity of material objects is that two observers have similar sensa, as opposed to a case of double vision, where one person has two sets of sensa; in the persistence of material objects it is that one observer has the recurrent or reversible series of sensa. (Actually, the best evidence of persistence would be that A sees the object during the gaps in B 's observation of it, for which mention of observers is clearly essential.)
  2. A more fundamental objection is that the assertion of the publicity and persistence of material objects is meant to convey more than the assertion of sets of sensa: One is maintaining, first, that a public object exists as the focus of two persons' perceptions and, second, that such an object continues to exist during the gaps between series of perceptions. ("Focus" here means either a common object of both perceptions, as in direct realism, or the common cause of the different sensa, as in representative realism.) It might be objected that this is simply putting forward an alternative to phenomenalism, but it seems fair to say that something like this realist claim is what we mean by a material object. Without the notion of focus or continuant, the agreement of different people's sensa or the recurrent sequences of one person's sensa are incredible series of coincidences. Why, for example, are such agreements so common in perception of objects but so rare in pains or dreams or imagery? Surely because there is something besides the sense experiences responsible for the agreement, namely, a common object or cause.
  3. Furthermore, the fragmentariness of our perception of an object is closely correlated with our own actions, as are Ayer's reversible series. If sensa of a table are replaced by sensa of the view outside the window, we must have moved our head and have looked out of the window; if we get sensa of the interior of the room after an hour's gap, we must have dozed off or have gone out and returned. This seems to show that the sensa are caused by continuing objects, the room and furniture; since the fragmentariness of our observation of these objects is explained by our actions, we do not have to assume that the objects are fragmentary as wellindeed, if they were, we should find them and their sensa appearing and disappearing without any action on our part, like the Cheshire Cat in Alice in Wonderland.

causal properties and processes

Any material object is thought to possess and to exercise many causal properties (its various powers to affect other objects by heat, propulsion, impact, pressure, chemical or electrical properties), and the concept of such an object may be claimed to involve them. They are so important that for many philosophers (for example, Price) they form the main stumbling block to the acceptance of phenomenalism, at least of the factual kind. Not only are these causal properties regularly exercised when the object is unobserved (fire still boils the water when the cook is not looking, beams still support the floor and roof even when quite hidden, and so on) but the properties and processes involved in the causation of perceptionthe events in the eyes and nerves of percipientsare also rarely if ever observed, and then only by scientists with special equipment. Thus, one may often perceive or experience the effects of unobserved causal properties; hence, actual sensa may be causally dependent on what are only possible oneswhich is absurd.

Followers of linguistic phenomenalism may claim to avoid this. The observed movement of the hands of a clock caused by unseen works inside it, for example, is not a case of actual sensa due to possible ones. What one should say, rather, is that sensa of hands moving are sensed, and if one were to get sensa of the back of a clock with the cover removed, one would get sensa of cog wheels and shafts moving; or, more generally, given sensa of the effect, then if certain other sensa occur, sensa of the cause would also occurSe, and if Sx, then Sc. It must be noted that such an analysis presupposes the Humean, or regularity, view of causation, in which all that a causal relation amounts to is that the "effect" has been observed regularly to follow the "cause" (C causes E means whenever C, then E )any conviction that the effect is brought about by some force in the cause that compels it to happen is mere superstition or is to be explained psychologically as the projection of our feeling of expectancy. However, this analysis will not satisfy those who maintain other theories of causation.

But even granting the regularity view, there is a special difficulty for phenomenalism. Presumably the "ifs" in the phenomenalist analyses are equivalent to "whenever" and themselves state regularities; whenever the floor board is taken up, one sees the beams supporting the floor. Hence, if causal relationship means no more than regularity or constant conjunction, the formula "Se , and if Sx , then Sc " amounts to "Se , and whenever Sx , then Sc " or "Se , and Sx causes Sc ." However, this expresses a causal relationship different from the original one; it concerns X and C rather than C and E, and, more important, expresses a relation between sensa, suggesting that one lot of sensa causes another. Indeed, this last conclusion must follow if nothing but sensory experiences exist. Thus, "The beam supports the floor" becomes "If (whenever) you have under-floor sensa, you have beam sensa," and hence, "Under-floor sensa cause beam sensa"which is far from the original. (This point applies with greater force to the causation of perceptions; the causal properties of the percipient's nervous system must be expressed in terms of the sensa of some other person entirelynamely, the physiologist, who can observe them.)

It has been objected that all this is unfair; the causal language belongs only to material-object language, and causal relations are between material objects and events, while in the sensum language and analysis they are expressed as equivalent correlations. However, according to the regularity view of causation there is no reason why the relevant sensa, which are events and are regularly correlated, should not be causally connected. Hence, the difficulty illustrated by "under-floor sensa cause beam sensa" still stands; it suggests that causal connections are more than relations of sensa, and thus that phenomenalism is false.

Quite apart from this special difficulty, the proposed analyses of causal properties are open to the general difficulties of the phenomenalist account of the existence of objects. There is a similar impurity, particularly with respect to the causation of sense experiences, analysis of which involves reference to different observers. Equivalent translation is even more clearly ruled out: Since causal properties involve other objects as well as the object analyzed, they are more complex than such simple, sensible ones as color or shape and thus require a longer and more intricate set of sensum statements for their analysis. They also produce their effect only when a whole range of standing conditions holds, for instance, the spring will not drive the clock if the bearings are clogged with dirt. All these conditions would have to be specified for the mutual entailment of a causal material-object statement and a set of sensum statements.

See also Perception; Sensa.


statements and defenses

In addition to Hume's own writings, of special interest are attempts to modernize and improve on Hume: Henry Habberley Price, Hume's Theory of the External World (Oxford: Clarendon Press, 1940), and Alfred Jules Ayer, The Foundations of Empirical Knowledge (New York: Macmillan, 1940).

On sensibilism, see Bertrand Russell, Mysticism and Logic (London: Allen and Unwin, 1918), and Price's Hume's Theory of the External World. Russell's theory developed from the phenomenalistic views he put forward in Our Knowledge of the External World (London: Allen and Unwin, 1914), which were criticized by C. D. Broad in "Phenomenalism," in PAS 15 (19141915): 227251.

Factual phenomenalism is expounded by John Stuart Mill in An Examination of Sir William Hamilton's Philosophy (London, 1872), Ch. 11 and appendix to Ch. 12. H. H. Price, Perception (London: Methuen, 1932), is sympathetic, though Price finally abandons factual phenomenalism. For allied views, see Karl Pearson, The Grammar of Science, 3rd ed. (London: A. and C. Black, 1911), or Ernst Mach, Contributions to the Analysis of Sensations (Chicago, 1897; London, 1900).

Linguistic phenomenalism, including criticisms of earlier variations, may be found in A. J. Ayer, The Foundations of Empirical Knowledge (see above), and "Phenomenalism," in PAS 47 (19461947): 163196, reprinted in his Philosophical Essays (London: Macmillan, 1952); however, his views in The Problem of Knowledge (London: Macmillan, 1956) involve some recantation. Also useful is D. G. C. MacNabb, "Phenomenalism," in PAS 41 (19401941): 6790. Compare the sophisticated version by Clarence Irving Lewis in his Analysis of Knowledge and Valuation (La Salle, IL: Open Court, 1947)comments on this in Roderick M. Chisholm, "The Problem of Empiricism," in Journal of Philosophy 45 (19) (1948): 512517; a reply by Lewis, ibid., 517524and Roderick Firth, "Radical Empiricism and Perceptual Relativity," in Philosophical Review 59 (1950): 164183 and 319331.

general criticisms and discussions

Introductory surveys are given by John Hospers, Introduction to Philosophical Analysis (New York: Prentice-Hall, 1953; London: Routledge and Paul, 1956), and Charles Harold Whiteley, Introduction to Metaphysics (London, 1955). There is a general discussion by A. C. Ewing, R. I. Aaron, and D. G. C. MacNabb in the symposium "The Causal Argument for Physical Objects," in PAS, Supp. 19 (1945): 32100.

More advanced and definitely critical are Rodney Julian Hirst, The Problems of Perception (London: Allen and Unwin, 1959); Alfred Cyril Ewing, Idealism, a Critical Survey (London: Methuen, 1934); David Malet Armstrong, Perception and the Physical World (London: Routledge and Kegan Paul, 1961); J. J. C. Smart, Philosophy and Scientific Realism (New York: Humanities Press, 1963); Wilfrid Sellars, Science, Perception and Reality (New York: Humanities Press, 1963), Ch. 3; Richard Bevan Braithwaite, "Propositions about Material Objects," in PAS 38 (19371938): 269290; R. I. Aaron, "How May Phenomenalism Be Refuted?" in PAS 39 (19381939): 167184; George Frederick Stout, "Phenomenalism," ibid., 118; Isaiah Berlin, "Empirical Propositions and Hypothetical Statements," in Mind 59 (235) (1950): 289312; W. F. R. Hardie, "The Paradox of Phenomenalism," in PAS 46 (19451946): 127154; and Paul Marhenke, "Phenomenalism," in Philosophical Analysis, edited by Max Black (Ithaca, NY: Cornell University Press, 1950).

other recommended titles

Adams, Robert Merrihew. "Phenomenalism and Corporeal Substance in Leibniz." Midwest Studies in Philosophy 8 (1983): 217258.

Ayer, A. J. Language, Truth and Logic. 2nd ed. New York: Dover, 1946.

Brandom, Robert. "Pragmatism, Phenomenalism, and Truth Talk." Midwest Studies in Philosophy 12 (1988): 7593.

Fumerton, Richard. Metaphysical and Epistemological Problems of Perception. Lincoln: University of Nebraska Press, 1985.

Graff, Delia. "Phenomenal Continua and the Sorites." Mind 110 (2001): 905935.

Jolley, Nicholas. "Leibniz and Phenomenalism." Studia Leibnitiana 18 (1986): 3851.

Kamooneh, Kaveh. "Hume: A Supernaturalist and a Phenomenalist." Philosophical Inquiry 24 (2002): 95102.

Moser, Paul K. "Beyond Realism and Idealism." Philosophia 23 (1994): 271288.

Sosa, Ernest. "Epistemology Today: A Perspective in Retrospect." Philosophical Studies 40 (1981): 309332.

Tye, Michael. "Phenomenal Consciousness: The Explanatory Gap as a Cognitive Illusion." Mind 108 (1999): 705725.

Tye, Michael. "A Theory of Phenomenal Concepts." In Minds and Persons: Royal Institute of Philosophy Supplement: 53, edited by Anthony O'Hear. Cambridge, U.K.: Cambridge University Press, 2003.

Van Cleve, James. "C. I. Lewis' Defense of Phenomenalism." Philosophy and Phenomenological Research 41 (1981): 325332.

Williamson, Timothy. Identity and Discrimination. Cambridge, MA: Blackwell, 1990.

R. J. Hirst (1967)

Bibliography updated by Benjamin Fiedor (2005)