1.  Central Concepts and Issues, Alternative Positions, and Important Arguments

         In this first section, I want to offer an overview of the issues that arise in this topic of perceptual knowledge of, or justified perceptual beliefs about, the external world.  In the first section, I shall set out what I take to be the main alternative positions that there are with regard to justified perceptual beliefs about, or perceptual knowledge of, the external world   Then, in section two, I shall list some important questions that one would like to be able to answer.  Finally, in section three, I shall briefly list some of the important arguments that bear upon the question of which of the main alternative positions on the justification of perceptual belief is the most plausible.

1.1  The Main Alternative Positions

         In outlining the main positions that have been advanced with respect to the issue of perceptual knowledge of, or justified beliefs about, the external world, I shall formulate the positions in terms of justified beliefs.  So for each position I list, there is a corresponding position that refers to knowledge, rather than to justified beliefs.  Keeping that in mind, one can then say that the main positions that have been advanced are as follows:

(1)  Solipsism/Skepticism

         This is the view that one has no justified beliefs about either the existence of, or the nature of, external objects.

(2)  Partial Skepticism with regard to External Objects

         This is the view that while one can have justified beliefs about the existence of external objects, one cannot have any justified beliefs about the nature of such objects, about their intrinsic properties.  (Compare Kant on the noumenal world, the world of things in themselves.  While Kant holds that one cannot have any knowledge of the intrinsic properties of noumenal objections, he seems to hold that one can at least be justified in postulating distinct noumenal objects, so one can know something about the noumenal world, beyond its mere existence.)

(3)  Skepticism with regard to Physical Objects

         This is the view that one has no justified beliefs about either the existence of, or the nature of, external, mind-independent, physical objects.  (Compare Berkeley's view.  Berkeley, however, is not skeptical here, since he thinks that one can offer good reasons for holding that there are no physical objects.)

         The next four positions are versions of direct realism, where this is the view that perception can provide one with non-inferentially justified beliefs about physical objects.  This general position comes, however, in versions which differ with regard to at least three important issues, namely: (i)  Does perception provide one with direct access to some physical properties - such as colors - and hence with justified beliefs concerning the intrinsic nature of such properties?  (ii)  Are sensory experiences intentional states - for example, nothing more than the acquisition of beliefs, or dispositions to have beliefs, about physical objects - or are they, on the contrary, non-intentional states?  (iii)  Do our non-inferentially justified perceptual beliefs about physical objects arise via a causal process that runs through sensory experiences?

         The different versions of direct realism that philosophers have advanced can be generated by asking the following questions:

(1)  Is the 'sensuous quality of redness' something that exists in the external world, as a property of the surface of physical objects?

         If one is a direct realist, and also gives a "Yes" answer to this question, one is embracing the following position:

(4)  Direct Realism:  "Naive", or "Pre-Scientific", or Manifest Image Version

         This is the view, accordingly, that perception can provide one with non-inferentially justified beliefs about physical objects, and that the properties that objects seem to have under favorable conditions of perception are properties that they in fact have.  So the "manifest image" of the external world is a true image.

         Almost all present-day direct realists would, however, give a negative answer to the question, "Is the 'sensuous quality of redness' something that exists in the external world, as a property of the surface of physical objects?"  When this is done, the next question to consider is this:

(2)  Is the 'sensuous quality of redness' something that exists inside one's mind, as a property of one's experiences?

         If one is a direct realist, and also gives a "Yes" answer to this question, one is embracing either the following position, or else John Searle's view - which is described below:

(5)  Direct Realism:  Sellars' Version

         This is the view, accordingly, that perception can provide one with non-inferentially justified beliefs about physical objects, but that the sensuous properties that objects seem to have are really properties inside the mind.  They are properties of experiences, not properties of physical objects. The "manifest image" of the external world must be rejected in favor of the scientific image.

         If, on the other hand, one gives a negative answer to the question, "Is the 'sensuous quality of redness' something that exists inside one's mind, as a property of one's experiences?", one is embracing the following version of direct realism:

(6)  Direct Realism:  Armstrong's Intentional States Version

         This is the view that perception can provide one with non-inferentially justified beliefs about physical objects, but that there are no sensuous properties to be found anywhere in the world - either as properties of external objects, or as properties of one's experiences.  For, as regards the former, external objects have only the properties that are ascribed to them by physics, while as regards the latter, sensory experiences are misconceived when thought of as involving any sort of direct grasp of a property, such as redness.  Experiences are correctly viewed as intentional states.  Having an experience of redness is, as a first approximation, nothing more than the acquiring of a certain belief - the belief, namely, that there is something red out there in the world.  The belief that there is something red out there, moreover, is neither the false belief that an object sometimes has, in addition to the properties ascribed to it by physics, the sensuous property of redness, nor the false belief that some external object has the power to produce experiences that have the sensuous property of redness in perceivers under appropriate circumstances.  It is rather the belief that there is an object that has a property with whose intrinsic nature one is not directly acquainted - a property one knows not what, except that it is a property that is shared by various things, such as ripe tomatoes and fire engines.

(7)  Direct Realism:  Searle's Experiences as Intentional States Version

         In the selection from John Searle's book, Intentionality, which is reprinted in Pojman's anthology, Searle puts forward a type of direct realism that does not seem to agree exactly with any of the preceding three positions.  Searle is very far from being as explicit as one would like about what his position is, and it would have helped if he had said how his position differs from other  formulations of direct realism.  As I interpret Searle, his view is very similar to that of Sellars, in that he holds that there are sensuous properties, but that these are not properties of physical objects.  But he seems to diverge from Sellars in holding, not that experiences cause beliefs about physical objects, but that experiential states themselves are also belief states, in view of their causal interconnections with desires and actions.  (The first appendix to this chapter contains a more detailed exposition of this interpretation of Searle.)

         Finally, there are three alternatives to direct realism.  All three of these alternatives share the view that no beliefs about external objects are non-inferentially justified.  They differ, however, with regard to whether beliefs about external objects can be justified.  According to the first two alternatives, beliefs about external, physical objects can be justified on the basis of justified beliefs about one's own present sensory experience, plus one's memories of past sensory experiences.  According to the third alternative, no beliefs about external, physical objects can be justified.  But although no beliefs about external physical objects can be justified, one is justified in talking as if there were physical objects, provided that one does not view oneself as claiming that the propositions in question are true.

         How do the first two alternatives differ, given that they agree that beliefs about external, physical objects can be inferentially justified on the basis of memory and sense experience?  The answer is that they differ with respect to the answers that they give to the question of the analysis of statements about physical objects.

(8)  Classical, or Reductionist Phenomenalism

         This alternative to direct realism involves two claims.  First, all justified beliefs about external objects are inferentially justified on the basis of justified beliefs about present and past sensory experiences.  Secondly, sentences about physical objects can be analyzed in terms of sentences - including counterfactual sentences - that refer to nothing other than minds and sensory experiences.  ("Physical objects are nothing over and above actual and possible sensory experiences.")

         Classical phenomenalism, in short, treats physical objects as theoretical entities - rather than as things that are "directly" or "immediately" observable - and offers a reductionist account of the meaning of statements referring to such theoretical entities.

(9)  Indirect Realism, or Causal Realism, or the Representative Theory of Perception

         This second alternative to direct realism involves two claims.  First, all justified beliefs about external objects are inferentially justified on the basis of justified beliefs about present and past sensory experiences.  Secondly, sentences about physical objects cannot be analyzed in terms of sentences that refer to nothing other than minds and sensory experiences.  Physical objects, rather than being reducible to minds and sensory experiences, are mind-independent objects that  cause sensory experiences.

(10)  Fictionalist, or Instrumentalist Phenomenalism

         This final alternative to direct realism agrees with the representative theory of perception in holding that, contrary to what is claimed by classical phenomenalism, statements about physical objects cannot be analysed in terms of statements about sense experiences, either actual or possible.  But it disagrees with the representative theory when the latter claims that one can offer a satisfactory, inferential justification of beliefs about physical objects.  Such beliefs cannot, according to fictionalist, or instrumentalist phenomenalism, be justified.

         This does not mean, however, that one should not employ the theory of physical objects.  One can use that theory, so long as one does not regard the relevant statements as true.  For one can view the theory as simply a useful fiction, as an effective instrument for making reliable predictions about what sense experiences one is likely to have under various circumstances.

         It is this instrumentalist version of phenomenalism that is defended by W. T. Stace in the selection entitled, "Science and the Physical World:  A Defense of Phenomenalism", which is contained in the Pojman anthology.  It is also the position that one would naturally arrive at if one, first, agreed with Bas van Fraassen's contention - in his book The Scientific Image, and elsewhere - that one is justified only in believing that theories are empirically adequate, and not that they are true, but secondly, held that van Fraassen's drawing of the line between pure observation statements and theoretical statements is epistemologically naive and indefensible, and that a satisfactory way of characterizing the class of pure observation statements will be one that excludes all statements about physical objects from that class.

1.2  Central Concepts and Issues

         The central questions in this area include at least the following:

(1a)  Can perception, possibly together with memory, provide one with justified beliefs concerning (i) the existence, and (ii) the nature of external objects - that is, of objects that are external to one's own mind?

(1b)  Can perception, possibly together with memory, provide one with knowledge about (i) the existence, and (ii) the nature of external objects - that is, of objects that are external to one's own mind?

(2a)  Can perception, possibly together with memory, provide one with justified beliefs concerning (i) the existence, and (ii) the nature of external physical objects - that is, external objects that are located in space, or space-time, and that are not mind-dependent, in the sense of not being dependent upon anyone's mind?

(2b)  Can perception, possibly together with memory, provide one with knowledge about (i) the existence, and (ii) the nature of external physical objects - that is, external objects that are located in space, or space-time, and that are not mind-dependent, in the sense of not being dependent upon anyone's mind?

Comment

         Compare Berkeley's view of the world.  Berkeley held that one could have knowledge of the existence of objects that are external in the sense of not being dependent upon one's own mind, but rejected the view that one could have knowledge of the existence of mind-independent objects.

(3a)  Can perception provide one with non-inferentially justified beliefs about external objects, or physical objects?

(3b)  Can perception provide one with non-inferential knowledge about external objects, or physical objects?

(4a)  If perception cannot provide one with non-inferentially justified beliefs about external objects, or physical objects, what sort of non-inferentially justified beliefs provide the basis of our justified beliefs about external objects, or physical objects?

(4b)  If perception cannot provide one with non-inferential knowledge about external objects, or physical objects, what sort of non-inferential knowledge provides the basis of our knowledge of external objects, or physical objects?

(5a)  If the most basic justified perceptual beliefs about external objects, or physical objects, are inferentially justified on the basis of beliefs about (past and present) sensory experience, what account can be given of the inference?  What reason is there for thinking that it is sound?

(5b)  If the most basic perceptual knowledge of external objects, or physical objects is inferential knowledge, and is based on non-inferential knowledge of present sensory experience, plus memory knowledge of past experiences, what account can be given of the inference?  What reason is there for thinking that it is sound?

(6)  Are all sentences about physical objects analyzable, or are some semantically basic, and unanalyzable?

(7)  If all sentences about physical objects are analyzable, what is the correct analysis?

(8)  Are all sentences that can be used to describe sensory experiences less basic than sentences about physical objects, or are there sentences about sensory experiences that are semantically basic, and unanalyzable?

(9)  If sentences about sensory experiences are analyzable, what is the correct analysis?  (Two main alternatives:  (i)  Armstrong's intentional account;  (ii)  Smart's topic neutral account.)

(10)  If there are sentences about sensory experience that are not analyzable, what is the logical grammar of such sentences?  (Two main alternatives:  (i)  A sense-datum view;  (ii)  An adverbial account.)

1.3   Important Issues and Arguments

        The five most important issues seem to me to be the following:

(1)  Do experiences involve emergent properties?

(2)  Is talk about physical objects analyzable, or, on the contrary, are at least some sentences about physical objects semantically basic and unanalyzable?

(3)  If all sentences about physical objects are analyzable, what is the correct analysis?

(4)  Does perception always involve the acquisition of beliefs about sense experiences?

(5)  If perception always involves the acquisition of beliefs about sense experiences, do the beliefs thus acquired, together with memory knowledge, suffice to justify the beliefs about physical objects that one comes to have as a result of perception?

        What are the main arguments that bear upon these five issues?  The answer seems to me to be as follows:

Issue 1:  Do Experiences Involve Emergent Properties?

        Here I think that there are eight arguments that are especially important:

(1)  Thomas Nagel's "What It's Like to Be a Bat" Argument;

(2)  Frank Jackson's "What Mary Doesn't Know" Argument;

(3)  The Inverted Spectrum Argument;

(4)  Armstrong's Competing Hypotheses Objection;

(5)  The Problem of the Relation between Mind and Body;

(6)  The Indeterminacy Objection;

(7)  Armstrong's Intransitivity Objection;

(8)  Armstrong's Epistemic Objection.

        The first three of these arguments support the claim that experiences involve emergent properties.  The last five arguments offer reasons against postulating such properties of experiences.

Issue 2:  Are Sentences About Physical Objects Analyzable?

        Here there are, I believe, two important arguments:

(1)  The Ostensive Definability Requirement;

(2)  The "Blind Man" Argument.

Issue 3:  If Sentences About Physical Objects Are Analyzable, What Is the Correct Analysis?

        The crucial arguments here, it seems to me, consist of three objections to phenomenalism:

(1)  "Mindless World" Objections;

(2)  The "Truth-Makers for Counterfactuals" Objection;

(3)  The "Exceptionless Laws" Argument.

Issue 4:  Does Perception that Results in Perceptual Belief Always Involve the Acquisition of Beliefs About Sense Experiences?

        Here I think that there are two important lines of thought:

(1)  The Peculiarity Intuition;

(2)  The Case of Abnormal Conditions of Observation.

Issue 5:  Can Beliefs About Physical Objects be Inferentially Justified?

        In the case of this final issue, I think that there are four important arguments:

(1)  The "Retreat to More Modest Beliefs" Argument;

(2)  The "Justification and Internal States" Argument;

(3)  The Appeal to Hypothetico-Deductive Inference;

(4)  The "Naturalness of the Theory of Physical Objects" Argument.
 

2.  Issue 1:  Do Experiences Involve Emergent Properties?

2.1  Argument 1:  Thomas Nagel's "What It's Like to Be a Bat" Argument

        Here is a brief exposition of Nagel's argument:

(1)  Bats are able to fly about without bumping into objects, even though they can't see, by sending out sound waves, and then perceiving those sound waves when they bounce off of nearby objects.

(2)  The sound waves that bats utilize fall outside the range of human hearing.

(3)  One could have, in principle, a complete anatomical and neurophysiological description of the relevant perceptual systems of a bat, so that everything that could be known about the physics of the situation could be known by us.

(4)  Nevertheless, mightn't one still wonder what the experience of a bat is like - given that the sound waves that bats perceive are ones that humans cannot perceive?  Mightn't one think that, even though one's anatomical and neurophysiological knowledge of bats was complete, there was still something that one did not know - namely, what the experience of a bat is like?

(5)  But if so, then there are facts about experiences that are not reducible to the facts that are within the domain of physics.  Those facts, however, are facts about the intrinsic nature of a bat's sonar experiences.  So there are properties of experiences that are not reducible to the properties that play a role in any of the theories in physics.

        How might one respond to Nagel's argument?  This question is best left until we have considered a second argument, advanced by Frank Jackson.

2.2  Argument 2:  Frank Jackson's "What Mary Doesn't Know" Argument

        Jackson's argument is as follows:

(1)  Mary is a person who has been born without color vision.  So she sees the world as black and white, plus shades of gray.

(2)  Mary goes on to study physics and neurophysiology, and comes to know all the facts that can be described in terms of physics, and that are relevant to colors, and to color perception.

(3)  If physics covers, in principle, all of reality, then it should be possible, in principle, for Mary to know all that there is to know about color, and color perception - or, at least, for her to know as much as you and I know.

(4)  But is this possible?  For suppose that Mary undergoes an operation that provides her with color vision, and she gazes, with new eyes, upon a ripe tomato.  Doesn't Mary, at that moment, acquire some new knowledge?  Doesn't she learn, for the first time, what redness is?

(5)  But if this is right, then there are propositions that one can know, but which cannot be reduced to propositions expressible in the language of physics.  And how can this be so unless there are properties that are not reducible to the properties that enter into theories in physics?

        How might one respond to Jackson's argument?  For isn't it true that Mary knows something afterwards that she didn't know before?  If so, what can that knowledge consist in other than knowledge of, for example, what redness is?

        One important response involves the idea that there are different kinds of knowledge.  In particular, there is propositional knowledge - or "knowledge that" - and there is knowing how to do something.  Given this distinction, some philosophers have argued that what Mary acquires when she sees something red for the first time is not any sort of propositional knowledge - including knowledge that that's what redness is - but, rather, knowledge of how to identify red things by means of one's eyes, rather than by means of, for example, devices that measure wave lengths of light.

2.3  Argument 3:  The Inverted Spectrum Argument

        One way of responding to the above response to Frank Jackson's argument involves an appeal to a third argument in support of the thesis that experiences involve emergent properties - the inverted spectrum argument.  This argument can be formulated in a number of slightly different ways.  One way of setting it out is as follows:

(1)  Consider the sort of experience you have when you look at something red, under normal conditions of illumination, etc.  It is different from the experience you have when you look at something green under similar conditions, providing that you are not red-green color-blind.  Now can't you imagine what it would be like for ripe tomatoes to give rise, not to the sort of visual experience you now have when you look at them, i.e., a sensation of redness, but instead to the sort of experience that you now have when you look at a ripe lime?  And conversely, can't you imagine what it would be like if the experience you got when you looked at a lime were like the experience you now get when you at a tomato?

        More generally, one can imagine what one's experiences would be like if one's complete visual spectrum were, so to speak, to be flipped over.  And not merely can one imagine this, one could make a film which would produce the sort of experiences one would be having if one's visual spectrum were to be flipped over.

        Moreover, this possibility is not merely a logical possibility.  It is at least empirically impossible, even if not technically possible at present.  For it will surely be possible some day to insert a device in a person that will systematically alter the messages sent from the eyes to the visual centers in the brain, and thus produce an inversion of one's visual spectrum.

(2)  The particular words that one uses to describe the colors of objects, and to describe one's visual experiences, surely do not affect the nature of those experiences.  You might, for example, decide to use the word "green" to describe objects that everyone else uses the word "red" to describe.  This would not affect the way that ripe tomatoes and limes look to one; this would not alter the visual experiences one has.

(3)  Suppose now that you have an identical twin.  In the interests of philosophy, your twin is separated from you, and placed in an artificial environment in which there are no natural objects such as tomatoes and limes which tend to be of fixed colors.  Your twin is then taught a language that is just like English, except that the use of color words is switched around. If you were to see an object which she would label "red", you would label it "green", and so on.

(4)  Suppose that her eyes and the visual portions of her brain are exactly similar to yours.  Then you have excellent reason to believe that the visual experiences she has when she looks at something which you would call "red" and she would call "green" are the same in quality as the experiences you would have when you look at the same object.  For it has been argued that the language one learns to describe visual experiences and the properties thereof does not significantly affect the quality of those experiences.

(5)  An operation is now performed on your twin which results in an inverting of her visual spectrum.  Objects that, before the operation, she would have labeled "red", she will now label "green".  Hence when you and she now look at the same object, your color experiences will be different from hers, just as your own color experiences would be different looking at a given object before and after such an operation.

(6) Your twin is now let out of her artificial environment.  The claim is now that, although when she looks at a ripe tomato she will have the sort of visual experience that you have when you look at a lime, there will be no relevant difference between you either with respect to actual behavior or with respect to behavioral dispositions.

(7)  Conclusion:  One can imagine cases, and in the future will actually be able to construct them, in which one has differences in mental states that are not accompanied by differences in behavioral states.  Hence mental states involve properties that are not reducible to behavioral states.

        The argument can equally well be expressed simply in terms of one's own case.  One way of expressing the argument is in terms of something that might have happened to you in the distant past.  Thus, you can imagine what it would be like for your spectrum to invert right now.  If so, you can equally well imagine what it would be like for your spectrum to have inverted at some time before you learned to speak, say, at the age of one month.  If that had in fact happened, you would be having a different experience when you looked at a ripe tomato from what you now have when you look at one.  But your behavior and behavioral dispositions might well be just the same as they are now.  Hence the properties of one's visual experiences cannot be identified with properties of behavioral states.

        Another way of putting the argument in terms of yourself focuses upon what could happen to you right now.  In this case, we need to consider three possible changes that might occur to you now.  One involves a change to your brain that inverts your visual spectrum.  A second involves a change to the linguistic center in your brain, so that your use of language for visual properties also undergoes a systematic inversion.  And finally, a third change is needed to some of the memories that you currently have - namely, memories concerned with your previous visual experiences and your use of language in relation to those experiences.  The idea is then to consider, first, making these changes one at a time, and then secondly, making the changes all at once.  When the latter is done, it seems very plausible to say that the experiences you have, for example, when you look at a ripe tomato will be different from what they were before the changes, even though, first, you will not be aware of the fact that they are different, and secondly, there will be no changes in your behavior or behavioral dispositions.

        The inverted spectrum argument just set out is also relevant to Frank Jackson's argument about what Mary does not know.  For if Mary's brain had been changed in certain ways before the operation that gave her the ability to see, then that would not have affected in any way the fact that after the operation she knew how to identify red things using her eyes, but it would have meant that her experience would have been different when she saw the first red object, and so she would have acquired a different belief  when she acquired the belief that this is what redness is.

2.4  Argument 4:  The Complexity Involved in Postulating Emergent Properties

        In his book A Materialist Theory of Mind, David Armstrong offered a number of interesting objections to the view that there are emergent, sensuous properties.  The thrust of one of these objections is that, if one does admit the existence of such properties, one's picture of the world will necessarily be much more complex.

        This increased complexity has a number of aspects:

(1)  One is postulating extra properties to deal with only a very small part of the world;

(2)  One needs to postulate a large number of emergent properties;

(3)  The causal laws that must be postulated are very different from the laws postulated by physics, and rather peculiar;

(4)  A large number of new laws need to be postulated.

(1)  Minds as Making Up Only a Very Small Part of the Universe

        Armstrong's first point is that the part of the spatiotemporal world that involves minds is extraordinary small.  For almost all of the physical universe, the type of account of reality that is offered by physics seems extremely good, and there is no reason at all for thinking that emergent, sensuous qualities are involved at any point.  How likely is it, then, that in that very small part of the spatiotemporal world that contains organisms with minds, physics will prove inadequate, and it will turn out to be necessary to postulate extra, emergent properties?

(2)  Qualia and the Postulation of a Large Number of Emergent Properties

        Given that humans can discriminate between quite a large number of visual, auditory, tactile, olfactory, etc., experiences, and that this is so even if one considers experiences of the very simplest sorts, it would seem that if one holds that experiences involve emergent properties, one will be committed to postulating, not just a few additional, basic properties, but a very large number of simple properties.  How does this compare with the picture of the world that one has if one agrees with Armstrong that physics provides, in principle, a complete description of the world?  In particular, how do the two views of the world compare with respect to the number of basic properties that they postulate?  To answer this question, one needs to consider whether, for example, the existence of objects with a large number of different masses is comparable to the existence of experiences with a large number of different, simple, color qualities.  If these were comparable, then the postulation of a large number of emergent, sensuous properties would not make for much of a difference with regard to complexity.  It seems to me, however, that they are not comparable.  The reason is that - leaving aside Hume's missing shade of blue case - it certainly seems to be true that one can understand what a particular sensuous property is like only if one has previously experienced an instance of that property, whereas one need not, for example, have encountered something with a mass of exactly n kilograms in order to know what it is for an object to have that specific mass.  So it would seem that there is a sense in which, in admitting emergent, sensuous properties, one is postulating many more basic, or simple, properties than are being postulated by physics.

(3)  The Peculiar Nature of the Causal Laws that Must Be Postulated

        Consider laws of Newtonian physics - such as Newton's Second Law of Motion - F = ma - or his Law of Gravitation - F = Gm1m2/r2.  Such laws involve quantitative properties, and they serve to relate an enormous number of properties falling under the relevant determinables.  Thus one does not need a different law specifying the acceleration of a body for each different combination of force and mass: a single law covers all cases.

        Compare this with the case of laws dealing with the causal dependence of qualia upon brain states.  Phenomenologically considered, qualia appear to be simple properties, not reducible to one another, nor to any simpler properties.  If this is right, then for every type of quale there will have to be a corresponding, basic law.  Such basic laws, moreover, will be very different from the sorts of laws postulated by physics, all of which connect very simple properties.  The laws that relate brain states to qualia, on the other hand, will be laws according to which extremely complex brain states, involving an enormous number of fundamental particles, causally give rise to experiences with simple properties.  Such causal connections between very complex properties and simple properties are laws of a very different sort from those that have been found to hold in the case of physics.

(4)  The Need for a Large Number of Extra Laws

        As was noted above, the fact that humans can discriminate between quite a large number of sensory experiences, of the simplest sorts, means that in admitting the existence of emergent, sensuous properties, one is committed to postulating a large number of simple properties.  This means, in turn, that one will need a large number of basic laws.  Moreover, the situation is even worse when one takes into account the fact that there is reason to believe that the sensory experiences of other species, with different perceptual systems, may very well involve many other sensuous properties.

        In short, while physics seeks out simpler and more comprehensive theories with fewer and fewer basic laws - aiming at the ideal of a single law - friends of qualia must hold, it seems, that when it comes to laws concerning how brain states give rise to experiences with qualitative properties, one needs to postulate an enormous number of basic laws.

2.5  Argument 5:  The Problem of the Relation between Mind and Body:  Epiphenomalism, Interactionism, or a Pre-established Harmony?

        A second, important objection that David Armstrong sets out in his book A Materialist Theory of Mind focuses upon the following question:  What causal connections, if any, are there between one's mind and one's body?  According to the pre-established harmony view, there are none at all.  What one has is the illusion of causal connections - an illusion generated by the fact that the mental and the physical started out in step, and then have remained in step as each changes in accordance with laws that hold within each realm.  But this view is surely implausible.  For it requires that there be causal laws linking physical events only to physical events, and mental events only to mental events, and while the former poses no problem - the laws in question are just those discovered by physics - there do not appear to be any laws linking mental events just to mental events: regardless of what experiences one has had at one time, one may have no experiences at all at some later time, due to injury to one's brain.

        One is left, then, with the choice between epiphenomenalism and interactionism, and neither seems very appetizing.  According to epiphenomenalism, states of consciousness are caused by brain states, but do not themselves have any causal power - except possibly with respect to other states of consciousness.  (An experience of the green variety may, compatibly with epiphenomenalism, give rise to the thought that one is having an experience of the green variety.)  But if this is right, then one is faced with what some writers have referred to as the paradox of epiphenomenalism - which is that everything in the physical world would be just as it presently is if there were no states of consciousness at all.  God, for example, could eliminate the laws linking brain states as causes to experiences as effects, and the physical world would be unchanged.  Human bodies would utter the same words as before, would continue (apparently) to talk about their experiences and thoughts and feelings, to debate about the nature of the mind, and so on.  Or God might replace the existing laws by new ones according to which certain brain states give rise to states of consciousness only when those brain states are found in female bodies.  No thoughts or feelings or other states of consciousness would then be associated with male bodies from that point on, and yet females would be unable to detect any difference.  Males would continue to appear to be the same deeply sensitive and caring creatures they have always seemed to be.

        This leaves, as the final option, interactionism.  This view has the advantage that our ordinary causal beliefs are preserved.  Thus, for example, when someone says that the grass looks green to them, that utterance is caused, at least in part, by the experience of the green variety to which the grass has given rise.  So far, so good.  But interactionism has implications that may be less attractive.  For it implies, not only that physics is not the complete picture of reality, but that physics is not even the complete picture with respect to fundamental particles, and things constituted out of fundamental particles and nothing else.  For if qualia can play a causal role in, for example, the utterances that people make, then various molecules must be moving differently due to the presence of certain states of consciousness than they would be moving in the absence of such states.  So if one wants to predict how air molecules, for example, are likely to be moving, one will do a better job if, rather than making use only of the laws of physics, one also employs the laws dealing with causal connections running from brain states to states of consciousness, and from states of consciousness to relevant brain states.

        Is this such a troubling conclusion?  Perhaps not.  But it is important to see exactly what it implies.  For the justification for believing in qualia, as things stand, is a matter of direct introspection.  So without leaving one's arm chair, one can have a justified belief in the existence of qualia.  But the arguments against a pre-established harmony, and against epiphenomenalism, are also essentially arm-chair arguments, involving at most quite low level empirical observations - such as that a knock on the head can render one unconscious.  Taken together, however, these arm-chair arguments enable one to conclude that when one applies physics to certain very complex things - namely, the brains of humans and other sentient creatures - one will find that the predictions are not as good as one would expect if the brains in question had only physicalistic properties.

2.6  Argument 6:  The Indeterminacy Objection

        In his earlier  book, Perception and the Physical World, Armstrong offered three arguments against the view that experiences involve special emergent properties.  The first of these three arguments runs as follows:

(1)  "Hume's Principle of the Transparency of Experience":

If Mary is having an experience with some qualitative property Q, then, necessarily, Mary will believe that she is having an experience with quality Q.

(3)  It follows from (2), given Hume's Principle, that in such a case there is no number n such there are exactly n black dots in Mary's visual field.  For:

(a)  Suppose that there is some number n such that there are exactly n black dots in Mary's visual field.

(b)  It then follows, given Hume's Principle, that Mary believes that there are exactly n black dots in Mary's visual field.

(c)  But this contradicts the premise advanced at (2), according to which there is no number n such that Mary believes that her visual field contains exactly n dots.

Indeterminate Properties?

        But why can't an object have indeterminate properties?  One answer is as follows.  What would it be for something to have an indeterminate property?  What would it be, for example, for something to be green, but not any particular shade of green?  One explanation is this.  X is green, but no particular shade of green, if and only if the following is the case:

(1) Either X is green 0 or X is green 1 or X is green 2 or ... or X is green n

- where green 0, green 1, green 2, ... green n represent all possible determinate shades of green

(2)  None of the following statements are true:  "X is green 0";  "X is green 1"; "X is green 2"; . . . "X is green n-1"; "X is green n".

        But this situation is logically impossible, for a disjunctive statement cannot be true unless at least one of its disjuncts is true - given the way that the logical connective "or" is defined in terms of truth-tables.

        The upshot is that unless there is some other way of explaining what it is to have an indeterminate property, we must conclude that indeterminate properties are logically impossible.  But it doesn't seem that there are any other explanations available.

        Consequently, the conclusion would seem to be that there aren't really such things as indeterminate properties out there in the world.  There are only indeterminate descriptions of things.  So there is not really a property of having more than thirty dots and less than forty dots.  That's a concept that may apply to something, but if it does apply, it applies in virtue of the object's having one of a number of completely determinate properties - such as the property of having exactly 31 dots, the property of having exactly 32 dots, ... the property of having exactly 39 dots.

(5)  Conclusion:  Experiences, rather than being extensional entities with emergent properties are instead intentional states - something like the acquisition of a belief.  For intentional states - such as beliefs - can perfectly well be characterized by "indeterminacy".  Mary can perfectly well believe that there are more than 30 black dots on a sheet of paper, and believe that there are less than 40, without believing, for some specific number n that there are exactly n black dots on the sheet of paper.

Response:  Hume's Principle should be rejected as false.

        But is this plausible, or simply ad hoc?  Answer: even if one thought that Hume's principle was true in the case of simple properties, why should one think that it was necessarily true in the case of complex properties?  For even if one is necessarily aware of simple properties, to get from awareness of simple properties to beliefs about complex properties - such as that of containing exactly 36 dots - may very well involve calculation, and calculation is surely neither compulsory nor immune to error.

2.7   Argument 7:  Armstrong's Intransitivity Objection

        Armstrong's second argument in Perception and the Physical World may be put as follows:

(1)  The emergent properties that experiences are supposed to have are supposed to be properties with which one is directly acquainted.  So they are observable properties.

(2)  Suppose that an experience involves at least two parts, one of which has an observable, emergent property, P, and the other of which has an observable, emergent property, Q - where P and Q may be the very same property.

(3)  Then it must surely be possible for the person having the experience to tell, by introspectively observing the experience, and without considering anything else, whether property P is the same property as property Q, or a different property.

(4)  If the person can detect a difference between property P and property Q, then property P must be a different property from property Q.  If, on the other hand, the person cannot detect any difference between property P and property Q, then - given that one is dealing here with observable properties of one's own experience - then property P must be the very same property as property Q.

(5)  It is possible for there to be three color samples - A, B, and C - that give rise to color experiences E, F, and G, respectively, such that (a) the person cannot discriminate between experiences of type E and experiences of type F, (b) the person cannot discriminate between experiences of type F and experiences of type G, but (c) the person can discriminate between experiences of type E and experiences of type G.

(6)  It follows from (4) and (5) that property E is identical with property F, and property F is identical with property G, but property E is not identical with property G.

(7)  But this is a contradiction.

(8)  Therefore the assumption that experiences involve emergent properties with which one can be directly acquainted must be rejected.

Possible Responses?

Response 1:  Quality of Experience Depends upon the Context

        The idea behind this first response is to challenge the idea that color samples of type A, B, and C give rise, respectively, to emergent properties E, F, and G regardless of what other color samples one is looking at on a given occasion.  So the suggestion is, in effect, that if one looks at color samples of types A and B, in the absence of a color sample of type C, then the result need not be an experience that involves properties E and F.  It might instead be an experience that involves two instances of the same property - perhaps some property intermediate between property E and property F.  Similarly, when one looks at color samples of type B and C at the same time, perhaps that results, not in an experience that involves properties F and G, but two instances of some other property that may be intermediate between F and G.  And if this is right, then the contradiction is undercut.

Comments

1.  Is this an ad hoc response?  That is to say, does this view have anything to recommend it other than the fact that it provides a possible solution?

2.  One question worth pressing is what happens when one looks at samples of types A, B, and C at the same time.  But perhaps that question isn't easy to answer on any view of the matter.

3.  It is very tempting to think that there could be universal laws linking brain states to emergent properties, some of which are not pairwise distinguishable.  (For why shouldn't "scanners" of emergent qualia be limited in their discriminatory capacities?)  If so, the first solution seems wrong.

Response 2:  Observational Knowledge is More Limited

        Direct acquaintance does not provide one with knowledge of exactly what property is present.  To be directly acquainted with a property may only provide one with the knowledge that one out of a range of properties is present.

2.8   Argument 8:  Armstrong's Epistemic Objection

        Armstrong's third argument in Perception and the Physical World might be put as follows:

(1)  If sense experiences involve emergent properties that one is directly aware of, then one can get from knowledge of such experiences to knowledge of physical objects only if either (i) phenomenalism, or (ii) the representative theory of perception is true.

(2)  Phenomenalism is false, and the representative theory of perception is false.

(3)  Therefore, the postulation that sense experiences involve emergent properties leads to the skeptical conclusion that one cannot have any knowledge of physical objects.

(4)  Therefore one should reject the view that sense experiences have emergent properties.

Two Responses:

Response 1:   This argument overlooks the possibility of a Sellars-style direct realism, according to which the occurrences of experiences with emergent properties is causally prior to beliefs about physical objects, but there are no beliefs about those experiences that are epistemically prior to beliefs about physical objects.

Response 2:   Either phenomenalism is true, or the representative theory of perception is true.

3.   Issue 2:  Are Sentences about Physical Objects Analyzable?

        The way to approach this question, I want to suggest, is by raising a very general issue:  Where should analysis start?  What terms should form the starting point for the analysis of other terms?  What terms are semantically basic?

        The meaning of many terms can be explained in terms of other terms.  But in the case some terms, such verbal definition does not seem helpful.  One needs to acquire an understanding of some terms by what is often referred to as "ostensive definition" - by being exposed to certain cases where the term is being actually used.

        But what counts as an "ostensive definition"?  One view is that one has an ostensive definition of a term where a person is perceptually acquainted with the property or relation in virtue of which the term in question applies to a situation.  As a characterization of ostensive definition, this is probably fine.  However, so understood, many terms that can be ostensively defined can also be defined verbally.  If there is to be a class of semantically basic terms in a strong sense, it will have to consist of terms that can only be defined ostensively.  Is there such a class?

        The idea behind this question can be brought out if one compares, say, the terms "kangaroo", or "duck-billed platypus" with the term "red".  The previous terms will be acquired ostensively by many people, but they can also be acquired via a purely verbal definition.  In this respect, they seem to differ from a term like "red".  For the meaning of the term "red", it might seem, cannot be conveyed via a purely verbal definition.

3.1  Can Physical Objects Terms Only Be Learned Ostensively?

        [This section is currently being revised.]
 

3.2   Talk about Physical Objects Versus Talk about Sense Experiences

        The arguments offered above may seem to support the conclusion that terms that are used to attribute observable properties to physical objects can in principle be understood in terms of terms that are used to attribute corresponding properties to sense experiences.  But one must be careful not to leap to that conclusion.  For suppose that one accepts the claims, first, that terms that are used to attribute sensuous properties to experiences can be ostensively defined only in circumstances where one has an experience that really has the relevant property, and secondly, that it is therefore appropriate to take such terms as semantically basic.  How can one get from sentences that involve such terms, such as

(1) Experience E is a red* experience (or: an experience of the red variety)

(2)  Physical object P looks red.

(3)  Physical object P is red.

        What about the move from (2) to (3)?  The key thing that is needed if sentences such as (3) are to be analysed in terms of sentences such as (2) is some account of normal conditions, and of a normal perceiver.  For the natural way in which to attempt to analyze (3) is as follows:

"Physical object P is red"   means the same as

"Physical object P would look red to a normal perceiver under normal conditions".

    One way of appreciating the difficulties that there may be even in the move from sentences such as (1) to sentences such as (2) is to consider how physical objects differ from sense experiences, and then to ask whether talk about the properties that characterize physical objects, but not sense experiences, can nevertheless be cashed out in terms of talk about sense experiences.

Sense Experiences Versus Physical Objects

(1)  Any sense experience belongs to a single individual, and can be "perceived" only by that person: sense experiences appear to be private.  Any physical object, by contrast, can be perceived by a variety of people:  physical objects are intersubjectively perceivable.

(2)  Sense experiences exist only as long as they are perceived.  Physical objects, by contrast, exist even when they are not perceived.

(3)  Sense experiences are mind-dependent:  a given sense experience would not exist if a certain sentient being did not exist.  Physical objects, on the other hand, are mind-independent.  For not only does the existence of a physical object not depend upon the existence of any particular mind; it does not, it would seem, depend on there being any minds at all.

        So if one is to analyze sentences about physical objects, it is not enough to explain the meaning that various predicates - such as "red" and "round" - have when they apply to physical objects: one also has to show that one can make sense of those crucial non-observational properties of physical objects that serve to distinguish physical objects from sense experiences.  In particular, one has to show that one can make sense of:  (1) being such as can be perceived by more than one person; (2) of existing even when not perceived; and (3) of existing in a way that is independent of the existence of any mind.

        Perhaps this can be done.  Certainly many philosophers in the past have felt that such an analysis of talk about physical objects in terms of talk about sense experience is possible.  My point here, however, is that one should be careful about the conclusions that one draws from the above arguments.  For although the arguments offered above do seem to show that terms attributing properties to experiences are in some respects more basic than terms attributing corresponding properties to physical objects, it has to be viewed as an open question whether talk about physical objects can be completely analyzed in terms of talk about sense experiences.

4.   Issue 3:  What is the Correct Analysis of Sentences about Physical Objects?

Classical Phenomenalism Versus the Representative Theory of Perception

        If sentences about physical objects can be analyzed, there are two alternatives: (1) a classical phenomenalist analysis; (2) the type of analysis offered by the representative theory of perception.

4.1  What Is Classical, or Reductive Phenomenalism?

        Recall that phenomenalism comes in two quite different forms.  On the one hand, there is classical, or reductive phenomenalism, and, on the other, there is fictionalist, or instrumentalist phenomenalism.  (It is the latter that is discussed and defended by W. T. Stace in the selection entitled, "Science and the Physical World:  A Defense of Phenomenalism", in the Pojman anthology.)  In what follows in this section, the focus is upon classical, or reductive phenomenalism.

        One way of characterizing classical phenomenalism is in terms of the claim that, although physical objects are certainly real, they are not basic entities, but, rather, what might be called "logical constructions" out of other entities.  In particular, the phenomenalist claims that physical objects are logical constructions out of sense experiences.  In referring to physical objects, and talking about them, one is not really referring to anything over and above sense experiences.  Physical objects are reducible to sense experiences.  (Analogy:  Armies and nations are certainly real, but an army is nothing over and above the soldiers that make it up, and a nation is nothing over and above the people who are interrelated in various ways, and, in being thus interrelated, constitute the nation.)

        But what does it mean to say that physical objects are logical constructions out of sense experiences, or that they are nothing over and above sense experiences?  There are two closely related, but slightly different ways of expressing this in a more precise way.  One way is to use the idea of analysis, and to define classical phenomenalism as follows:

Classical Phenomenalism is the view that:

(1)  All sentences about physical objects can be analysed in terms of sentences that refer to nothing other than minds and their experiences and other mental states.

(2)  All justified beliefs about physical objects are inferentially justified on the basis of beliefs about sense experiences.

        This is a very natural formulation.  However some philosophers contended that there were reasons for doubting whether such an analysis could be carried out - at least, without employing infinitely complex sentences about sense experiences.  Accordingly, an alternative account emerged that, rather than claiming that sentences about physical objects could be translated into sentences about sense experiences, appealed to an apparently weaker relation - the relation of logical supervenience.

        Precisely how supervenience is best defined is not entirely clear, but for present purposes the following will do.  Let S and T be any two set of sentences that are related in such a way that if the truth-values of all of the sentences in set T are fixed, then the truth-values of the sentences in set S must also all be determined.  Then the sentences in set S are logically supervenient upon the sentences in set T.

        Given this notion of logical supervenience, classical phenomenalism can be characterized in the following way:

Classical Phenomenalism is the view that:

(1)  The set of all sentences about physical objects is logically supervenient upon the set consisting of all sentences that refer to nothing other than minds and their experiences and other mental states.

(2)  All justified beliefs about physical objects are inferentially justified on the basis of beliefs about sense experiences.

4.1.1   Reductive Phenomenalism:  The Underlying Idea

        So much for a characterization of classical, or reductive phenomenalism.  But to get a sense of the basic idea that lies behind this version of phenomenalism, it may be helpful, I think, to consider how such a phenomenalist attempts to explain the difference between really seeing something, on the one hand, and, say, merely imagining that one is seeing something.  One difference, typically, is that one's experiences are much more vivid in the normal case of seeing than in the normal case of merely imagining that one is seeing something.  But there are other features that phenomenalists would generally view as more important, such as:

(1)  Though one can control one's perceptual experiences in certain ways - e.g., by closing one's eyes, or looking in a different direction - one does not have the sort of voluntary control over one's perceptual experiences that one has over the content of what one is imagining.

(2)  Another aspect of this difference with respect to voluntary control is that one's perceptual experiences involve very frequent correlations between different sorts of perceptual experiences - such as visual experiences and tactile experiences - correlations that need not exist in the case of imagined experiences.  Thus if one sees something that looks round, that thing will almost invariably feel round as well - if one touches it - whereas one can perfectly well imagine something that looks round, but that feels square.

(3)  Perception of objects is characterized by a certain intersubjectivity that does not typically characterize imagined experiences.  If I see three red balls lying on a table, then others in my vicinity will tend to have similar visual experiences, whereas what I am imagining may be completely different from what others around me are imagining.

        One idea that lies at the heart of classical phenomenalism, in short, is that genuine, veridical perceptual experiences exhibit certain patterns and correlations, both in the case of a single person, and intersubjectively, and it is precisely these patterns and correlations that mark the experiences out as perceptual experiences of an external reality, and that distinguish them from mere imaginings and other purely subjective experiences.

4.1.2   Classical Phenomenalism:  Actual Versus Hypothetical Experiences

        Can we, then, view physical objects as nothing over and above the actual sense experiences that people have, as nothing more than, say, certain patterns in the actual sense experiences of people?  This view encounters the problem of the tree in the quad:  What is one to say about the tree when nobody is present in the quad?  If the tree is nothing over and above the actual sense experiences that people have, and which we would normally describe as sense experiences "of" the tree, then mustn't we conclude that the tree does not exist when there is no one looking at it?

But physical objects are, by definition, things that can exist unperceived.  If this is right, then we cannot identify physical objects with collections of actual sense experiences.  But how, then, can physical objects be viewed as logical constructions out of sense experiences?  The phenomenalist's answer is that, even when no one is in a quad, there is a difference between a quad that contains a tree and one that does not - and a difference, moreover, that can be explained in terms of sense experiences.  For if a quad contains a tree, then, even if no one is observing the tree right now, it will be true that if someone were in the quad, they would be having tree-type sense experiences, whereas this would not be true in the case of a quad that does not contain a tree.

        The phenomenalist's idea, then, is that one needs to refer not merely to the actual experiences that people have, but also to the experiences that people would have if things were different in various ways.  So subjunctive conditionals concerning possible experiences need to be brought into it.  Physical objects, while not reducible to actual experiences, can be reduced to the combination of actual and hypothetical experiences - where talk about hypothetical experiences is to be cashed out in terms of subjunctive conditionals, in terms of talk about what experiences people would have under various circumstances.

        In short, one might imagine starting from the following attempt to explain what it is for there to be, say, a tree in the quad:

(1)  "There is a tree in the quad" means the same as "People have experiences of the tree-in-the-quad variety"

        But that's obviously inadequate, since it provides no account of the existence of the tree when no one is perceiving it, let alone of the tree in the quad that no one ever perceives.  To get around this, reference to hypothetical experiences are introduced.  But how should that be done?  Suppose one offered the following:

(2)  "There is a tree in the quad" means the same as "People have experiences of the tree-in-the-quad variety, and, also, if anyone were in the quad at an appropriate time, they would have experiences of the tree-in-the-quad variety."

        The problem with (2) is that while the subjunctive conditional that is introduced to "refer" to hypothetical experiences does not refer to the tree, it does refer to another physical object - namely, the quad.  So this sort of analysis will not do if all statements about physical objects are to be analysed in terms of sentences about sense experiences.  But (2) can be improved upon:

(3)  "There is a tree in the quad" means the same as "People have experiences of the tree-in-the-quad variety, and, also, if anyone were to have an experience of the being-in-the-quad variety at an appropriate time, they would also have experiences of the tree-in-the-quad variety."

        But this isn't quite right either, since there are quite a variety of experiences of the being-in-the-quad variety, some of which would not be, so to speak, experiences of being in the right quad.  Consequently, the being-in-the-quad experiences that are referred to in the antecedent of the subjunctive conditional need to be described very precisely, so that the right quad, so to speak, is picked out.  It is clear, moreover, that that will be a very complex matter, given that there are many different locations that one can occupy in the quad, each giving rise to different experiences.  So one needs something like this:

(4)  "There is a tree in the quad" means the same as "People have experiences of the tree-in-the-quad variety, and, also, if anyone were to have an experience of a certain complex, being-in-a-quad variety at an appropriate time, they would also have experiences of the tree-in-the-quad variety."

        A final point to note is that one might well think that the reference to actual experiences should be dropped, and that one should say, if one is a phenomenalist, that physical objects are reducible to hypothetical experiences alone.  For if one brings in actual experiences, then it seems that what a given object consists of varies, depending upon when someone has had an experience of it.  If one did this, the analysis that one would be left with would be along the following lines:

(5)  "There is a tree in the quad" means the same as "If anyone were to have an experience of a certain complex, being-in-a-quad variety at an appropriate time, they would also have experiences of the tree-in-the-quad variety."

4.2  Objections to Classical, or Reductive Phenomenalism

4.2.1  Armstrong's Objections to Classical Phenomenalism

        Armstrong, in his discussion of phenomenalism in his book, Perception and the Physical World, offers the following six objections to classical, or reductive phenomenalism:

(1)  Phenomenalism entails that unperceived physical objects have only a "hypothetical" existence;

(2)  Phenomenalism entails that a universe that contained no minds would contain no matter either;

(3)  Physical objects are determinate, and cannot be constructed out of indeterminate sense experiences;

(4)  Phenomenalism cannot explain the public nature of space and time;

(5)  Phenomenalism can give no account of the numerical difference of different minds that exist at the same time;

(6)  Phenomenalism cannot provide a satisfactory account of the nature of a mind.

Objection 1:  Phenomenalism Entails that Unperceived Physical Objects Have Only a "Hypothetical" Existence

        Armstrong's first objection is one that is not as clear as it might be.  Phenomenalism, as we have seen, does cash out talk about unperceived physical objects in terms of hypothetical, subjunctive conditional statements about the experiences that perceivers would have if they were differently situated.  So the existence of an unobserved tree in the quad is, according to phenomenalism, a matter of the truth of hypothetical statements.  But it seems misleading to talk, as Armstrong does, about "hypothetical existence", for the latter phrase suggests that one is dealing with something that doesn't really exist, as things stand, but which would exist if things were different in certain ways.

        Compare, here, the case of a sugar cube that is not presently in water.  If it were in water it would dissolve.  Because of that, there is a hypothetical event that one can refer to - the event of the sugar cube's dissolving at a certain time.  Such an event is purely hypothetical, if the cube is not actually in water at that time.  But the sugar cube also has a certain property - the property of being water-soluble - and that property is a real property.

        It seems to me, then, that Armstrong is in effect failing to distinguish between (1) hypothetical events - such as the sugar cube's dissolving in water at a time when it is not in fact in water - and (2) states of affairs involving dispositional properties of objects - such as the sugar cube's being water-soluble.  Hypothetical events do not, it is true, correspond to anything actual, but states of affairs involving dispositional properties of an object are perfectly real.  If this is right, then the phenomenalist's response to Armstrong's first objection is that the hypothetical statements in terms of which the phenomenalist explains the existence of unperceived objects do correspond to actual states of affairs, just as real dispositional properties correspond to the relevant subjunctive conditionals in the case of things like the water-solubility of a piece of sugar.

Objection 2:  Phenomenalism Entails that a Universe that Contained No Minds Would Contain No Matter Either

        This second objection of Armstrong's is, I think, one of the most interesting, though I don't that think it is sound as stated.  As Armstrong develops it, it runs as follows:

(1)  The phenomenalist analyzes talk about unperceived physical objects in terms of talk about hypothetical experiences - that is, talk about the experiences that perceivers would have if circumstances were different in certain ways.

(2)  If subjunctive conditional statements are to be true, there must be some relevant universal, lawlike generalization that is true.  (So, for example, if it is to be true that this particular sugar cube would dissolve if it were placed in water, that must be because there is some relevant, universal, lawlike generalization that is true - such as the generalization that all pieces of sugar dissolve when placed in water.)

(3)  The phenomenalist, in attempting to show that it is logically possible for there to be a world that contains physical objects, but no minds, will have to bring in subjunctive conditionals of the form "If there were minds, then if there were certain experiences of such and such a type, there would be other experiences of some other type."

(4)  Such a subjunctive conditional can only be true if there is some appropriate universal, lawlike generalization concerning minds and experiences.

(5)  A lawlike generalization cannot be true unless either there are actual positive instances of it, or else it follows from some other, stronger, lawlike generalization for which that is true.

(6)  In the present case, that would have to be a generalization about minds and/or experiences.

(7)  Such a generalization can be true only if there are minds and/or experiences,

(8)  In the world being imagined, there are no minds, and so no experiences either.  Therefore, no relevant generalization could be true in that world, from which it follows that no relevant subjunctive conditional could be true.

(9)  Conclusion:  Phenomenalism entails that it is logically impossible for there to be a world that contains physical objects, but no minds.

Comments

1.  Armstrong's argument presupposes that there cannot be laws that lack any positive instances.  But there are strong reasons for rejecting that claim - e.g., (a) the case of two types of fundamental particles that never meet, and (b) the case of psychophysical laws in a  world where life is destroyed just before the primordial Snoopy sees his first purple flower.

2.  But I think that Armstrong's argument can be revised to avoid this objection.  The key to doing so is the realization, first, that subjunctive conditionals often are made true by a combination of two different sorts of facts - the existence of a relevant law, together with the existence of a relevant categorical state of affairs - and secondly, that the subjunctive conditionals to which the phenomenalist must appeal are subjunctive conditionals of this latter sort, rather than subjunctive conditionals that are made true by laws alone.  For imagine a world that contains both a round, red ball, and a round, green ball.  If there were a sentient being in the vicinity of the first ball, what would be true would be that if one were to have certain tactile experiences of the round-feeling variety, then one would have certain visual experiences of the round and red variety, whereas if a sentient being were instead in the vicinity of the second ball, what would be true would be that if one were to have certain tactile experiences of the round-feeling variety, then one would have certain visual experiences of the round and green variety.  So what subjunctive conditionals would be true would not always be the same, and this means that the subjunctive conditionals in question can't be made true by laws alone.  The truth of different subjunctive conditionals will require appropriate categorical facts.  (Compare what makes it true that if this object, X, were in water, it would dissolve, whereas if some other object, Y, were, in water, it would not dissolve.  It is a matter of both laws and categorical facts about the different molecular structures of X and Y.)

Objection 3:  Physical Objects are Determinate, and Cannot be Constructed out of Indeterminate Sense Experiences

        The thrust of Armstrong's third objection is as follows:

(1)  Physical objects are completely determinate.

(2)  Sense experiences are indeterminate.

(3)  There is no satisfactory way of constructing determinate physical objects out of indeterminate sense experiences.

        The argument for the claim made at step (3) is not entirely clear cut, and one might very well challenge it.  But I think that the phenomenalist is better advised to reject the claim that sense experiences are indeterminate - a claim that, as we have seen earlier, rests upon an appeal to Hume's Principle.

Objection 4:  Phenomenalism Cannot Explain the Public Nature of Space and Time

        In claiming that physical objects are logical constructions out of sense experiences, the phenomenalist is not saying that physical objects are logical constructions simply out of his or her own sense experiences.  The idea is that physical objects are logical constructions out of everyone's sense experiences.  However, the phenomenalist cannot arrive at this public conception of physical objects in a single step.  The reason is that, initially, only one's own experiences are given, and one must establish that there are experiences belonging to other people before one can hold that there are physical objects viewed as constructs out of everyone's sense experiences.

        The phenomenalist must, in short, proceed as follows:

(1)  One establishes that there are objects that are logical constructions out of one's own sense experiences, and that are external in the sense of existing even when one is not perceiving them.

(2)  Having established that there are external objects in that limited sense, one goes on to argue that there are sense experiences associated with some of those external objects, and that those sense experiences do not belong to oneself.

(3)  Having established that there are other minds, one must then go on to establish that, for each such other mind, there is an external world that consists of objects that are logical constructions out of the sense experiences that belong to that mind.

(4)  Finally, one can argue that there is a world of public physical objects that is defined in terms of relations between the individual external worlds that one has already constructed.

        Armstrong's fourth objection focuses, in effect, upon the final step here, and he claims that no adequate account can be given of that step.  In essence, I think that his objection involves three main claims.  First, he argues that the fact that Mary's external world resembles John's external world does not suffice to justify the claim that there is a public world that both John and Mary are experiencing.  (He appeals to the analogy of people in two different movie theaters, watching the same movie.)  Secondly, he argues that the combination of resemblance of structure plus simultaneity of resembling parts isn't sufficient either.  (People could be watching the same movie in different theaters at the same time.)  Thirdly, he argues that the phenomenalist, starting out from the idea that the experience of a single individual may involve different aspects which are simultaneous, cannot arrive at any satisfactory account of what it means to say that some experience that John is having is simultaneous with some experience that Mary is having.  So the phenomenalist cannot offer anything beyond resemblance in attempting to unite individual external worlds into a public world, and resemblance is not adequate.

Comments

1.  Armstrong's argument is interesting, but I think that he fails to consider one of the most plausible suggestions concerning what it is that unites individual external worlds into a single public world.  That suggestion is that it is counterfactual dependence that plays the crucial role.

Suppose, for example, that John is having an experience of the green apple variety, and so is Mary.  What makes it the case that they are having experiences of one and the same public object - a certain green apple?  A natural answer, I think, is this.  Suppose that Mary decides to paint the town red, including the external green apple that inhabits her external world.  It turns out that when she does this, the external green apple that inhabits John's external world also gets painted red.  Then I think that there is reason for saying that the two, formerly green apples that inhabited John's world and Mary's world constitute a single public object.

Of course, Mary may not in fact paint the town red.  But the idea then is that a certain counterfactual may nevertheless be true - namely, that if she had painted her apple red, then John's apple would have undergone a similar change of color - and that it is the truth of such counterfactuals that makes it the case that John's external world, and Mary's external world, are part of one and the same public world.

2.  But although Armstrong has overlooked a crucial response to his challenge, may it not be that he has an adequate answer to that response?  For if one appeals to relations of counterfactual dependence between John's experiences of his external world and Mary's experiences of her external world, then one is confronted with the question of what makes the relevant subjunctive conditionals true.  What makes it true, for example, that if Mary's present apple-experience had been of the red variety, rather than of the green variety, then John's present apple-experience would also have been of the red variety, rather than of the green variety?

        In thinking about this question, notice that a crucial feature of such counterfactuals is that they are counterfactuals linking the experiences of different people, and, because of this, it would seem that any categorical facts that make such counterfactuals true will have to be facts that involve the experiences of both people.  But then one is back, it seems, with Armstrong's original challenge, since the relations that come to mind which are categorical relations - in contrast to the relation of counterfactual dependence, which has to be explicated in terms of subjunctive conditionals - don't seem adequate.

        But I think this is to move too quickly.  The reason is that what makes the subjunctive conditionals true need not be a matter of a direct categorical link between the experiences.  For suppose that the phenomenalist is able, first, to construct a world of objects that are external relative to his or her own experiences, and also that he or she can establish laws linking brain states with experiences.  Then if your descriptions of your experiences match up with the experiences that it would be reasonable for me to expect you to have, given the nature of that external object that is your body, and the way in which it is being stimulated in my external world, then the hypothesis that the experiences thus produced cause your verbal behavior enable me to explain why your external world - that is, the world that could be constructed out of your experiences - maps into my external world.

        To illustrate, suppose that my external world contains a green apple.  Light striking that apple, and reflected on to the eyes of the external object that is your body, should, according to the relevant psychophysical laws, give rise to experiences of the round and green variety.  If they do, then the world that you construct on the basis of those experiences should be a world that contains a green apple.  If the green apple object in my external world now ripens, or is painted red, then the same sort of reasoning will enable me to predict that your experiences will change accordingly, and thus that the external world that you construct will now contain a red apple, in place of a green one.

        The view, in short, is that the counterfactual dependence of your external world upon my external world results from the fact that your experiences are causally dependent, in view of the relevant laws, including laws linking brain states with experiences, upon external objects in my world.

        So it looks as if there may well be a satisfactory answer to Armstrong's fourth objection.  But one might feel that the resulting picture is peculiar in a certain way.  For consider your experiences.  The explanation just offered makes use of the claim that your experiences are causally dependent upon an external object in my external world - an object "constructed out of" my sense experiences.  But your experiences are also causally dependent upon an object in your external world - an object constructed out of your sense experiences.  So it looks as if your sense experiences are causally overdetermined - and indeed radically so, since they are also caused by the external objects in everyone else's external world!

    What one would like to do is to identify all of these external objects, and say that the experiences are caused by a single physical object.  But it would seem that in order to do this, one needs to move on to the representative theory of perception.  For if one treats a public object as a logical construct out of external objects in the external worlds of different people, then any person's sense experiences will still be causally overdetermined, since they will be caused by a large number of different "parts" of the relevant public object.  So perhaps there is a residual problem here that the phenomenalist cannot really handle in a satisfactory way.

Objection 5:  Phenomenalism Can Give No Account of the Numerical Difference of Different Minds that Exist at the Same Time

        Armstrong's fifth objection to phenomenalism may be put as follows:

(1)  It is logically possible for there to be two minds, existing at a given time, that are indistinguishable with respect to their mental states.

(2)  Can one distinguish the minds in question by reference to the bodies with which they are associated?  Armstrong says that, if phenomenalism is true, one cannot do this;  "We cannot appeal to the numerical differences of bodies in any way, for the bodies dissolve into sense-impressions which are part of the experiences of minds. There seems to be no criterion for numerical difference available."  (Perception and the Physical World, page 69)

(3)  Nor, Armstrong claims, will it do to appeal to the idea that numerical difference can obtain in cases where no further explanation can be given of how it is that two things are different.  Armstrong's view here seems to be that if one is to be able to hold that A and B are different things, then one must be able to provide some ground of that difference.  (In the case of physical objects, Armstrong's view is that two indistinguishable physical objects can nevertheless be two numerically distinct objects since they may occupy different spatial locations.)

(4)  Hence, if phenomenalism were true, it would be logically impossible for there to be two numerically distinct, but qualitatively indistinguishable minds.

(5)  But this is not impossible.

(6)  Therefore phenomenalism is false.

Comments

1.  I do not see how the claim advanced at (2) is to be justified.  For so long as bodies do in fact exist, why can't the phenomenalist appeal to them to individuate minds?  What difference does it make if physical objects are themselves logical constructions out of sense experiences?

        Suppose, for example, that one starts from one's own case, and constructs an external world out of one's own sense experiences.  One then uses the argument from analogy, say, to support the conclusion that there are other minds associated with certain sorts of physical objects.  Suppose that it turns out that there are two bodies that are, as regards their brains, physically indistinguishable at some time t.  Why wouldn't one be justified in concluding, in such a case, that there were two indistinguishable minds, one associated with each of the bodies in question?

2.  If one holds that physical objects are to be individuated on the basis of the fact that they have different spatial locations, the question rises as to how spatial locations are to be individuated.  Not, presumably, on the basis of having different intrinsic properties.  Nor will one be able to appeal to relations to other spatial locations, since all spatial locations may have the same properties.  Finally, if one tries appealing to different relations to physical objects, one will either wind up in a circle, if one appeals to particular physical objects, or, if one appeals to types of physical objects, one will run up against the standard counterexamples to the Principle of the Identity of Indiscernibles.

Objection 6:  Phenomenalism Cannot Provide a Satisfactory Account of the Nature of a Mind

        Armstrong advances three claims in this sixth objection:

(1)  If physical objects are logical constructions out of sense experiences, then, since sense experiences also belong to minds, it follows that minds and physical objects overlap, that they involve common elements, and this seems strange.

(2)  The phenomenalist cannot give any satisfactory account of the "uniting principle" of a mind - of what makes it the case that two mental states belong to one and the same mind.  (Armstrong seems to think that one needs to appeal to a body in order to have a satisfactory account.  Contrast his apparent view in his later book, A Materialist Theory of the Mind, where he says that nonembodied minds are possible.)

(3)  The phenomenalist has to accept, in the end, the "bundle theory" of the mind.  But the bundle theory leads to the absurd conclusion that any given sense experience could exist on its own, and not as a part of any bundle.  For a sense experience that existed on its own would be an unsensed sense experience, and that is unintelligible.

4.2.2 The Crucial Objections to Classical, or Reductive Phenomenalism

        The most fundamental objections to classical phenomenalism seem to me to be the following:

(1)  "Mindless World" Objections;

(2)  The "Truth-Makers for Counterfactuals" Objection;

(3)  The Exceptionless Laws Objection.

(1)  "Mindless World" Objections

        The first sort of objection comes in slightly different forms.  David Armstrong's second objection to phenomenalism is one version of a "mindless world" objection, and the thrust of it was that if phenomenalism is true, then a world without minds would also have to be a world without physical objects.

        Armstrong's own argument was, I argued, somewhat problematic, since it appealed to the assumption that basic laws must always have at least one positive instance.  However I also argued that subjunctive conditionals come into two different types - those which are entailed by laws, and those which obtain in virtue of laws together with categorical states of affairs.  Given this distinction, Armstrong's argument can be revised, since one can argue, first, that the subjunctive conditionals to which the phenomenalist must appeal in explaining the hypothetical experiences that there are in such a world are of the second sort, and secondly, that in a world without minds, and so without experiences, even if there are appropriate laws, there are no relevant categorical states to provide the relevant truth-makers for the subjunctive conditional statements in question.

        A second version of a "mindless world" objection appeals not merely to the possibility of a world where there happen not be any minds, but where there are no laws connecting brain states with states of consciousness, and so where the world is such that, no matter what the arrangement of physical objects is in the world, no minds will ever arise.

        The point about this second sort of mindless world is that it will not contain either relevant lawsor relevant categorical states of affairs which are necessary if one is to have truth-makers for subjunctive conditional statements concerning the experiences one would have if circumstances were different.

(2)  The "Truth-Makers for Counterfactuals" Objection

        This second objection turns upon the same basic idea as the "mindless worlds" objection, but the thrust of it is that the difficulty to which the "mindless worlds" objection points is a much more general one, in that it arises very frequently even in a world that contains minds.

        The basic idea underlying the extension is that in a world with minds - such as our own - there may very well be many objects that are never perceived, and many other objects that, though perceived, have aspects or properties that are never perceived, and the claim is that the same problem will arise for such objects, or aspects of objects, as arises for all objects in a world without minds - namely, that there will be no categorical states of affairs to serve as truth-makers for the subjunctive conditionals that characterize the relevant hypothetical experiences.  For although there are minds, those minds will not have any experiences that are related to objects that are unperceived, or to unperceived aspects of objects that are perceived.

(3)  The Exceptionless Laws Objection

        The thrust of this third objection is that there are no exceptionless laws dealing with sense experiences, since whether one has experiences at a given time, and precisely what experiences one does have, if one does have experiences, depends upon precisely what physical objects there are, and how they are related to one's body.  If there are to be exceptionless laws, one must postulate physical objects, and the exceptionless law-statements that result will then be statements that refer to the properties of, and relations among, physical objects.

        The argument here is most easily appreciated, I think, if one considers a closely parallel argument dealing with the reasonableness of postulating objects - such as atoms, or small molecules - that are too small for humans to perceive.  Consider Newton's First Law of Motion, which says that any body at rest remains at rest, and any body in motion continues in motion in a straight line with constant speed unless acted upon by some external force.  (It may also say more than this - namely that the velocity of any object remains constant unless it is acted upon by external forces whose vector sum is non-zero.)  Now consider Brownian motion, where very small particles that are observable - such as pollen, or scrapings from a piece of chalk - are present in a liquid, and are seen to move about constantly in an irregular fashion.  Since such particles are not moving with uniform velocity, if Newton's First Law of Motion is to be true, then there must be appropriate external forces acting upon them.  There are various hypotheses that might be advanced concerning those external forces, but it turns out that when those hypotheses are carefully considered, the only one that seems acceptable is that the particles that are moving about in a jerky fashion are doing so because of collisions with other particles that are too small to be perceived.

        In short, Newton's First Law of Motion can be sustained only if one postulates the existence of particles that, because of their size, are too small to be observed.  The same is true, however, with respect to other laws.  So either our world contains very small, unobservable things, or it does not have any laws - or at least, any exceptionless laws.

        Could one reject the idea of postulating unobservable objects in favor of either (1) holding that there are no laws at all, or (2) holding that there are only statistical laws?  The problem with the former is that it would leave one with massive cosmic coincidences - correlations of events that are, in the absence of relevant laws, extremely improbable.  The idea of merely statistical laws, on the other hand, would avoid the problem of cosmic coincidences, but it would fail to explain, for example, why objects of different sizes diverge, to different degrees, from the behavior they would exhibit if Newton's Laws of Motion were true.  Why is it, for example, that small objects move in a very jerky fashion when not acted upon, as it is being assumed, by any external force, whereas large objects appear to conform very closely to Newton's laws?  In contrast, if one postulates the existence of molecules, then that hypothesis, in conjunction with detailed assumptions about the sizes of molecules, their masses, and their kinetic energies will enable one to give a full account of Brownian motion.  One will be able to predict, for example, the osmotic pressure that will arise due to Brownian motion.

        In short, the argument, in the case of physical objects that are unobservable because of their size is this:

(1)  Small unobservable objects must be postulated if the world is to contain exceptionless laws.

(2)  A world with no laws at all would be a world containing cosmic coincidences - and, thus, events that would be extremely improbable.

(3)  A world with only statistical laws would be a world where there would also be unexplained cosmic coincidences - for example, ones dealing with the extent to which objects of various sizes, in different circumstances, diverge from the predictions that would be generated by exceptionless law-statements.  (Why do small objects move in a jerky fashion, whereas large objects, not acted upon by any external force, do not?)

(4)  Therefore the postulation of very small, unobservable objects is necessary if the occurrence of very improbable, cosmic coincidences is to be avoided.

        What about the postulation of macroscopic objects?  Here there are two different ways to go.  First, there is the reductionist route:

(1)  One is justified, in view of the preceding argument, of postulating the existence of very small, unobservable objects.

(2)  The detailed theories concerning the nature of such objects, and the ways in which they interact, entail that collections of such objects will have the power to act upon one another in ways that will give rise, in appropriately constituted collections, to sensory experiences.

(3)  Macroscopic objects can therefore be identified with certain very large collections of very small, unobservable objects.

(4)  Therefore we are justified in believing in the existence of macroscopic objects.

        The other possibility is an argument for the postulation of macroscopic physical objects that is similar to that given in the case of submicroscopic objects.  For it would seem that one wants to be able to old that people were justified in believing in the existence of macroscopic objects before they had any theory at all of the submicroscopic nature of reality.  The problem, however, is a theory that postulates macroscopic objects alone cannot both be true, and yield exceptionless laws.  But perhaps one can recast the above argument in terms of approximations to exceptionless laws:

(1)  Macroscopic physical objects must be postulated if the world is to contain even approximations to exceptionless laws.

(2)  A world without even approximations to exceptionless laws would be a world containing cosmic coincidences - and, thus, events that would be extremely improbable.

(3)  A world with only statistical generalizations about sense experiences would also be a world where there would be unexplained cosmic coincidences.

(4)  Therefore the postulation of macroscopic physical objects is necessary if the occurrence of very improbable, cosmic coincidences is to be avoided.

4.3  Beyond Phenomenalism:  The Representative Theory of Perception

        Why did phenomenalism once appear very attractive to many philosophers?  In the case of classical, or reductive phenomenalism, I think that there were two main sorts of considerations, the first semantical, and the second epistemological.  The former is inapplicable in the case of the other version of phenomenalism - fictionalist, or instrumentalist phenomenalism - but the epistemological consideration, in contrast, then becomes crucial.

4.3.1  The Semantical Support for Classical Phenomenalism

        Central to reductive phenomenalism are the ideas, first, that the only things that one is directly or immediately acquainted with are one's own experiences, and secondly, that one's justified beliefs about physical objects are justified via an inference from justified beliefs about one's own experiences.

        The semantical question is now as follows.  If the only things with which one is directly acquainted are one's own experiences, how can one understand what is meant by a physical object?  One is not directly aware of any physical objects, nor directly aware of any properties of physical objects.  So how can one ever understand the concept of a physical object, or the concept of some property - such as redness - that is a property of physical objects?

        The simplest answer, and the one that the phenomenalist accepts, is that physical object concepts can be analyzed in terms of concepts that apply to one's experiences.  Talk about physical objects is talk about something that is reducible to experiences.  In particular sentences about physical objects can be analyzed in terms of subjunctive conditional sentences about sense experiences.

        If one is to move beyond phenomenalism, one needs some other account of the meaning of sentences about physical objects.  One other answer that can be given, of course, is that one is directly aware of physical objects, and it is this direct awareness that enables one to understand the concept of a physical object, and the concepts of specific properties of physical objects.  But if one rejects this view, and holds that all of one's justified beliefs about physical objects are based upon justified beliefs about sense experiences, what alternative is there to the phenomenalist answer?

        If one is to abandon classical phenomenalism in favor of the representative theory of perception, one needs an answer to that question.

4.3.2  The Epistemological Support for Phenomenalism

        The epistemological argument for phenomenalism - which applies in the case of fictionalist, or instrumentalist phenomenalism, as well as in the case of classical, or reductive phenomenalism - involves an argument for skepticism that I sketched some time ago.  The thrust of that argument was, first, that the only legitimate forms of reasoning are (1) deduction and (2) induction via generalization from instances, and secondly, that statements about physical objects can neither be deduced from statements about sense experiences, nor arrived at via a process of generalization from instances.  In the case of deduction, the skeptic's argument appealed to possibilities such as evil demons and brains in vats, while, in the case of induction, the argument was that if one is directly acquainted only with sense experiences, then the only properties that one can be aware of as being found together in a given case will be properties of experiences, and so any generalization that one arrives at will be a generalization about properties of experiences: reference to the properties of physical objects will never come in at any point.

        If this argument is right, then it would seem that the only way to avoid skepticism with regard to physical objects is to hold that statements about physical objects can somehow be analyzed in terms of statements about sense experiences.  One must, in short, embrace phenomenalism.

        The alternative is to reject the argument.  One way of doing so, once again, is to embrace direct realism, and to hold that one is sometimes directly acquainted with physical objects.  If one doesn't want to accept direct realism, is there any other alternative?  If the representative theory of perception is to be defensible, some other way of responding to the argument must be found.  But what could that be?

4.4  The Representative Theory of Perception and the Two Challenges

4.4.1  The Semantical Challenge

        If one is directly acquainted only with one's own experiences, how can one ever form the concept of something that is not reducible to experiences?  The answer that an advocate of the representative theory of perception must give is, roughly, that one can form concepts that are general in a certain way.

        To make clear what I have in mind here, consider the following sequence of questions:

(1)  Can I form the concept of an experience that is not my own?

        The only experiences that I am aware of are my own experiences.  So how can I ever form the concept of an experience that is not my own?  For it is not the sort of thing that I can ever be directly aware of, and it is clear that talk about the experiences of other people can't be analyzed in terms of talk about my own experiences: the experiences of other people are not logical constructions out of my own experiences.

        A. J. Ayer in Language, Truth, and Logic accepted this conclusion, and held that when one used sentences that apparently referred to experiences of others, one was really just talking about the behavior that other bodies were exhibiting, and their behavioral dispositions.  Ayer was a behaviorist with regard to other minds, but not with regard to his own mind.

        What I want to say is that, starting from experiences, all of which belong to me, I can form the general concept of an experience, where that general concept is not such as entails that anything to which the concept applies must be an experience that belongs to me.

        One argument that I would offer in support of this claim involves an account of the truth conditions of sentences containing indexicals.  Consider, for example, the following two sentences:

(1)  This is a chair.

(2)  There is a chair.

        The idea now is to argue that sentences such as "I am now having an experience of type E" are also sentences that involve indexicals: the term "I' functions to pick out a specific person.  But if that's right, then an understanding of a sentence such as "I am now having an experience of type E" should enable one to understand sentences such as "Someone is now having an experience of type E".  (Some philosophers would prefer to start, not from sentences such as "I am now having an experience of type E", which involves the notion of a mind, or person, or subject of experiences, but from sentences of a simpler sort, such as "Here now an experience of type E", or "This is an experience of type E".  In that case the corresponding indexical-free sentence would be "There exists an experience of type E".)

        But if I can form such a general concept of an experience, then I can use it, together with the concept of an experience being mine, together with the concepts of negation and identity, to form the concept of an experience that does not belong to me, the concept of an experience that is someone else's.

(2)  Can I form the general concept of a property - a concept that covers properties that I may never have been directly aware of?

        I have experiences of a visual sort that involve direct awareness of various color properties.  Hume asked whether one could imagine a missing shade of blue.  That is not my question.  My question is whether, on the basis of acquaintance with various color properties, one can form the general concept of a color property, so that one will be able to understand a sentence like: "There may be color properties that I have never been directly acquainted with"

        But it is not the general concept of a color property that I'm really interested in here.  It is, rather, the perfectly general concept of a property.  And what I want to maintain is that just as it is plausible to think that one can form the general concept of a color property, so it is plausible to think that one can form the perfectly general concept of a property.

(3)  Can I form a general concept of existence - a concept that applies to experiences, but that may also apply to things that are not experiences?

        Again, my feeling is that just as I can conceive of experiences that are not my own, even though the only experiences I can ever be directly acquainted with necessarily belong to me, and just as I can form the concept of a property that is not restricted in its application to properties that I have been aware of at some time, so I can separate the idea of being real, or the idea of existing, off from the idea of an experience, even though the only real things that I have ever been directly acquainted with are experiences, and, thereby, I can form a concept of existence that may cover things other than experiences.

        But if I can do this, then the door is open for an account of the meaning of sentences about physical objects that does not reduce physical objects to anything else, including sense experiences, actual and hypothetical.  For I can begin by saying that a physical object is something that exists, and that is not an experience.  Of course, much more work will need to be done, in order to bring in other essential properties of physical objects.  One will, for example, need to bring in concepts of space and time.  One will also need to bring in causal notions, both in order to specify how physical objects are related to sense experiences, and to capture the crucial idea that the existence of physical objects is not causally dependent upon the existence of any mind.  But once one has the concept of something which exists, and which is not an experience, it seems plausible that the hurdles that remain will not be insurmountable:  reductionism is behind one, and it looks like a matter of working out the details.

        To sum up, then, what is the representative theorist's response to the semantical challenge?  In outline, the idea is that there is an alternative both to the reductive phenomenalist account, according to which statements about physical objects are to be analysed in terms of statements about sense experiences, and to the direct realist approach that treats at least some physical object terms as semantically basic.  The alternative involves the claim that terms of the following types can be taken as semantically basic:

(1)  Terms about experiences, the properties of experiences, and the relations within experiences - including, crucially, spatial relations;

(2)  Logical terms, such as 'and', 'or, 'not', 'all', 'some', 'there exists', 'identity', etc.;

(3)  Quasi-logical, or topic-neutral terms, such as 'property', 'relation', 'event', 'state of affairs', 'entity';

(4)  Nomological and causal terms.  (Alternatively, one could offer an analysis of nomological and causal terms.)

        If terms of these four types are available, then a statement such as:

"There are physical objects"

"There are entities that (1) are not identical with either minds or experiences, (2) stand in causal and spatiotemporal relations to one another, and (3) can causally give rise to experiences."

        This line of thought suggests, however, a slightly different alternative.  Perhaps one should add a fourth clause which is compatible with the above type of theism, and which says that physical objects are entities that: (4*) are not causally dependent upon the existence of any finite (or embodied?) minds?

4.4.2  The Epistemological Challenge

        What about the epistemological hurdle?  What the advocate of the representative theory of perception must argue here is that generalization based upon instances is not the only legitimate form of inductive reasoning.  The claim will be that one can also employ what has variously been described as "hypothetico-deductive method", or "the method of hypothesis", or "inference to the best explanation".

        One way of supporting this claim is by noticing that, however generous a view one takes of the range of objects with which one is directly acquainted, one must appeal to another type of inductive reasoning in order to justify beliefs about the physical world that are both very fundamental, and very widely accepted.  For one is certainly not directly aware of protons, neutrons, and electrons, let alone things such as quarks.

        It is, of course, possible to adopt a reductionist view of the talk, in physics, about sub-atomic particles.  But there are reasons for thinking that such reductionism won't, in the end, work, and if that is right, the only alternative to skepticism will be to appeal to some form of reasoning that is neither deductive, nor a matter of generalization from instances.  So the advocate of the representative theory of perception, in appealing to another form of inductive reasoning, is not engaged in an ad hoc exercise:  what he or she needs in order to set out a certain account of the justification of our beliefs about ordinary, macroscopic objects also appears necessary if one is to justify familiar beliefs about sub-microscopic objects.

5.   Issue 4:  Does Perception that Results in Perceptual Belief Always Involve the Acquisition of Beliefs About Sense Experiences?

        Here I think that there are two important lines of thought:

(1)  The Peculiarity Intuition;

(2)  The Case of Abnormal Conditions of Observation.

       [This section is currently being revised.]
 

5.2  The Case of Abnormal Conditions of Observation

        [This section is currently being revised.]
 

6.   Issue 5:  Are Beliefs About Physical Objects Inferentially Justified on the Basis of Beliefs about Sense Experiences?

        I think that there are four important considerations that are relevant to the view that justified beliefs about physical objects are always inferentially justified beliefs, based ultimately upon beliefs about sense experiences - both present experiences and past experiences.  The first two provide reasons for holding that the theories that accept this claim - namely, phenomenalism and the representative theory of perception - are preferable, other things being equal, to any form of direct realism that accepts the existence of phenomenal states, while the other two arguments serve to dispose of important objections to the claim that beliefs about physical objects cannot be non-inferentially justified.

(1)  The "Justification and Internal States" Argument;

(2)  The "Retreat to More Modest Beliefs" Argument - Or, Who's Afraid of Unconscious Inferences?

(3)  The Appeal to Hypothetico-Deductive Inference;

(4)  The "Naturalness of the Theory of Physical Objects" Argument.

6.1  The "Justification and Internal States" Argument

        [This section is currently being revised.]
 

6.2  The "Retreat to More Modest Beliefs" Argument.   Or, Who's Afraid of Unconscious Inferences?

        The basic strategy behind this argument is to bring out the extent to which the ordinary beliefs that we acquire as a result of perception can be viewed as involving unconscious inference.  The initial thrust of the argument is a defensive one - namely, to show that the fact that the representative theorist has to appeal to the idea that ordinary perception involves unconscious inferences does not constitute any objection to the representative theory of perception.  But the argument can also be used to add support to the representative theory, in that the final step that the representative theorist holds is necessary if one is to arrive at the relevant non-inferentially justified beliefs is just a final step in a series of steps that the direct realist must also take, all of which involve a move in the direction of a more modest description of one's perceptual experiences.

Illustrations

(1)  Examples of other sensory modalities than the one actually involved in the sense experience.  Berkeley's example of seeing a red, hot poker.  If one is some distance away, one doesn't feel the heat.  The belief that it is hot is an inference based upon one's memory of previous experiences.  Similarly, seeing a crisp, juicy apple.

(2)  Perceptual beliefs about a whole object based upon perception of a part of the object.  Seeing an orange versus seeing the front half of an orange: one's belief that there is a whole orange there may be based upon one's memories, of having seen it from a different perspective, or of having handled it, etc.

(3)  How things are versus how things look.  If one knew that the lighting conditions were abnormal, one would not believe that the object was red: one would believe only that it looked red.  Question:  In circumstances in which one does conclude that the object is red, is one unconsciously making use of the assumption that lighting conditions are normal?

Underlying Strategies

        There are, I think, two sorts of strategies that get appealed to here.  The more fundamental, I think, is this:

(1)  In some cases, such as Berkeley's case of seeing the red, hot poker, the appeal is to what beliefs one would have formed if one had not previously had certain experiences.  So, if one had never experienced hot objects previously, one would not have acquired the belief that one was seeing a red, hot poker.

        But another strategy is this:

(2)  One attempts to find some other statement that could be used to describe one's perceptual experience just as well as the statement that one naturally uses, but which  makes a more modest claim.  Thus, in the case of the statement, "I see an orange", one might introduce the statement, "I see what may be either half an orange, or the front of a whole orange", or, alternatively, "I see a hemispherically shaped piece of orange skin".  One then asks what justifies one in using the statement that involves the greater commitment, and tries to show how one's acceptance of the stronger statement may plausibly be thought to rest upon additional information that one has.

The Argument from the Logical Possibility of Complete Hallucination:  The Final Retreat

        Up to a point, the retreat to more modest perceptual claims will not take one beyond claims regarding physical objects.  But the strategy of the phenomenalist, or the advocate of the representative theory, is then to point to the possibility of complete hallucination, and to ask what the justification is for describing one's perceptual experiences in a way that entails the existence of a physical object, rather than in a way that is neutral between perception of a physical object, and a completely hallucinatory experience.

        The direct realist will want to resist the final step, even though it seems very similar to earlier steps.  I don't think that there is any principled way of doing so - that is, that any plausible grounds can be offered for holding that the only unconscious inferences involved in perception are inferences that take one from some beliefs about physical objects to other beliefs about physical objects, and that there are none that take one from beliefs about one's experiences to beliefs about physical objects.  But regardless of whether a principled way of resisting the final step is possible, the demonstration that perception involves an enormous amount of unconscious inference surely shows that it cannot be a serious objection to the representative theory of perception that it postulates unconscious inferences.

6.3  The Appeal to Hypothetico-Deductive Inference

        A traditional objection to the Representative Theory of Perception is that it makes it impossible for one to justify beliefs about physical objects.  The answer to this objection is that this is not so, provided that one is not confined to induction via instantial generalization, and that one can appeal to hypothetico-deductive method.

6.3.1  Is Hypothetico-Deductive Method Acceptable?

        Ultimately, one would like to be able to offer a justification of hypothetico-deductive method, to show that it is a legitimate way of supporting beliefs.  Unfortunately, there is not yet any agreement that a satisfactory solution to the general problem of induction is at hand.   There are, however, at least three points that should be noticed.

        The first is that, given that hypothetico-deductive method is crucial to the justification of scientific theories dealing with things that are too small to be observed, anyone who accepts a realist view of theoretical entities in physics can accept the representative theory of perception without appealing to any method of justification beyond those that one already needs.

        Secondly, notice that the fact that there is not yet available a satisfactory solution of the problem of induction is certainly not a reason for accepting induction via generalization upon instances, while rejecting hypothetico-deductive method.  For Hume's formulation of the problem of induction is specifically concerned with instantial generalization.

        Thirdly, a number of philosophers have recently argued that if laws were mere regularities, we could never be justified in believing that any exceptionless laws obtained.  For laws, so conceived, would be massive cosmic coincidences.  If we are to have reason for holding that there are laws, laws must be conceived as something other than regularities - for example, as second-order relations between universals.  But if laws, rather than being identical with regularities, are second-order states of affairs that explain regularities, then the existence of laws cannot be established via enumerative induction, by instantial generalization: the method of hypothesis will be needed.  So if the latter is unsound, no inductive inferences at all are possible.

6.3.2  Hypothetico-Deductive Method and the Representative Theory of Perception

        What will the theory be like that, according to the representative theory of perception, can be justified via hypothetico-deductive method?  It would seem to involve at least the following three elements:

(1)  Laws concerning the behavior of physical objects;

(2)  Laws concerning the causal connections between physical states and experiences;

(3)  Particular claims about the existence and arrangement of physical objects.

        Given statements of these three types, the idea is that one will be able to deduce conclusions concerning what sorts of experiences one will have, and if one does in fact have those experiences, then the theory in question is at least partially confirmed.

6.4  The "Naturalness of the Theory of Physical Objects" Argument

       [This section is currently being revised.]

Appendix 1:  John Searle's Version of Direct Realism

1.  Questions that Need to Be Answered Concerning Searle's View

Searle's overall view is not easy to pin down, because he fails to address directly certain crucial issues:

(1)  Does reality contain sensuous properties?

        Since Searle disagrees with those who reject visual experiences, and holds that cases such as blind sight show that visual experience cannot be reduced to the acquisition of beliefs, it is tempting to think that he holds that in visual experience one is aware of sensuous properties.  But he does not make any clear cut affirmation on this matter.

(2)   What properties do physical objects have?  Do they have sensuous properties?  Is sensuous yellowness, for example, a property of the surface of some station wagons?  If not, what does it mean for a station wagon to be yellow? Is it a dispositional property related to experiences with certain phenomenal properties?

        Searle says that it is a category mistake to apply a term such as "yellow" to experiences, and that it is only physical objects that are literally "yellow".  But unfortunately, he does not say what he thinks one is saying when one says that a station wagon is yellow.  So it's not clear whether he believes there are sensuous properties, and if so whether they characterize physical objects or parts of experiences.

        He also claims that one has direct access to physical sates of affairs, but again he does not explain what this claim comes to.

(3)  What are the phenomenal properties of experiences, and of parts of experiences?  Are they sensuous properties?  If not, what are they?

        Searle talks about its being a category mistaken to apply terms such as "yellow", or terms for shapes, to experiences, but unfortunately he does not indicate which terms can be applied to experiences - other than terms that apply to experiences as a whole, such as "pleasant".  The crucial question, however, is whether there are properties that are properties of parts of experiences, and if so, what they are.

(4)  Do experiences involve two aspects or parts - a phenomenal part and another part that carries the intentional content?

        My impression is that Searle holds that the phenomenal part "realizes", or carries, the intentional content, rather than there being two parts.  For he seems to hold that sameness of intentional content entails sameness of the experience, and thus - presumably - sameness of the phenomenal properties of the experience.

(5)  What exactly is involved in the "direct access" claim?  Is it simply a matter of non-inferentially justified beliefs, or does it involve direct acquaintance with properties?  If the latter, what are the properties that one is directly acquainted with?  Are they sensuous properties?  And how are they related to the phenomenal properties of experiences?

        Searle's discussion here is too brief to be at all helpful.

2.  A Possible Reconstruction of Searle's View

        The position that I'm inclined to ascribe to Searle, and my reasons, are as follows:

1.  Experiences do involve phenomenal properties

Reason:  Searle explicitly says this.

2.  The phenomenal properties of experiences are sensuous properties.

Reason:  The ground for this claim is much weaker.  But given Searle's rejection of certain positions in philosophy of mind, it seems unlikely that he would reject sensuous properties.  And it is very implausible to locate sensuous properties in the external world, given physics.  So, by default, they are properties of experiences and parts of experiences.

3.  Experiences are not composed of a mixture of phenomenal properties and propositional states.  Rather, since concepts are defined by their roles, it is possible for phenomenal states to have the roles in question, and this is so in the case of experiences.

Reason:  Searle seems to accept the inference:  Same intentional content, same phenomenal type of experience.

4.  The intentional content that phenomenal states realize involves propositions about physical states of affairs having powers to produce experiences of the very sort that realizes the intentional content in question.
 

3.  A Comparison with Other Versions of Direct Realism

        If the above reconstruction is right, then Searle's view differs from the other three versions of direct realism as follows:

1.  Unlike manifest image direct realism, Searle does not attribute sensuous properties to physical objects.

2.  Unlike Armstrong, but in agreement with Sellars, he does hold that there are experiences, with sensuous properties.

3.  He diverges from Sellars in not holding that beliefs about physical objects are caused by experiences with phenomenal properties.  Instead, he maintains that experiences themselves, in virtue of their phenomenal properties, and the causal relations into which those experiences enter, realize the relevant intentional states - specifically, beliefs about physical objects.
 

Appendix 2:  Flow Charts for Theories of Perceptual Knowledge
 
 

1.  Flow Chart for the Justification of Beliefs about Physical Objects

(1) Do "sensuous" properties exist, anywhere in reality?

Yes                           ª                                Question 2

No                            ª                                Armstrong's Version of Direct Realism
 

(2)  Do physical objects have the "sensuous" properties they seem to have?

Yes                           ª                                Naive, or "Pre-Scientific," Direct Realism

No                            ª                                Question 3
 
 

(3)  Can some beliefs about physical objects be non-inferentially justified?

Yes                          ª                                Sellars', or Searle's' Version of Direct Realism

No                           ª                                Question 4
 

(4)  Can beliefs about physical objects be inferentially justified on the basis of beliefs about sense experiences, past and present?

Yes                         ª                                 Classical, Reductive Phenomenalism, or
                                                                      the Representative Theory of Perception.
                                                                      Question 6

No                          ª                                Skepticism or Fictionalist Phenomenalism
                                                                      Question 5
 

(5) Even if beliefs about physical objects cannot be justified in the sense of being show to be likely to be true, is one justified in making use of such beliefs?

Yes                          ª                                Fictionalist Phenomenalism

No                           ª                                Skepticism
 

(6)  Can beliefs about physical objects be inferentially justified without using hypothetico-deductive method (inference to the best explanation)?

Yes                         ª                                 Classical, Reductive Phenomenalism

No                          ª                                 Either the Representative Theory of Perception,
                                                                      or else Classical, Reductive Phenomenalism
 
 

2.  Flow Chart for the Meaning of Statements about Physical Objects
 

(1)  Are any statements about physical objects basic and unanalyzable?

Yes                          ª                                Some Form of Direct Realism.

No                           ª                                Question 2
 
 

(2)  Are statements about physical objects analyzable in terms of statements about sense experiences?

Yes                           ª                                Classical, Reductive Phenomenalism

No                            ª                                Question 3
 
 

(3)  Are statements about physical objects analyzable in the way that statements about theoretical objects are analyzable?

Yes                          ª                                 The Representative Theory of Perception,
                                                                       or  Fictionalist Phenomenalism

No                           ª                                Question 1