Phenomenal Intentionality and the BIV

PHENOMENAL INTENTIONALITY AND THE BRAIN IN A VAT

George Graham, John Tienson, and Terry Horgan

In the First Meditation, Descartes envisioned the possibility that he was the victim of an enormously powerful, enormously clever deceiver, that his experiences were radically nonveridical, and that his beliefs were massively false. Versions of this scenario have figured centrally in epistemology ever since, especially in discussions of radical Cartesian skepticism. In recent philosophy, often the scenario is posed in a high-tech guise: as some variant of the “brain-in-a-vat” scenario.

Suppose, for instance, that there are some intelligent creatures elsewhere in the cosmos, who are very different from humans in physical composition and physical appearance, and whose surrounding environment is quite different from our own. They are in no causal contact with earth, and know nothing of it or its inhabitants. In the course of their scientific investigations, they synthesize a structure out of organic molecules that happens to be an exact physical duplicate of your own brain; they hook it to a computer in such a way that its ongoing brain-activity happens to exactly match your own, throughout its existence. Also, there is substantial counterfactual exact physical similarity between this artificial brain and yours; various counterfactuals about what would happen to and within your brain if certain physical events were to occur to it or within it—for instance, counterfactuals about what would happen physically if the brain were to receive certain potential physical inputs to its sensory-input neurons—are also true of the synthesized brain. Hereafter when we speak of the Brain in a Vat (for short, the BIV), usually we will mean this particular synthesized brain in this particular setup—a specific version of the generic brain-in-vat scenario.

When one first contemplates the generic scenario, certain pre-theoretic intuitive judgments about it tend to arise strongly and spontaneously, given the knowledge that physical processes underlying human mentality occur in the human brain. One intuitively judges that the BIV’s mental life exactly matches one’s own, in a very strong way; this means, inter alia, that the BIV has numerous beliefs, both perceptual and non-perceptual, that exactly match one’s own beliefs. Also, one intuitively judges that many of these matching beliefs—including perceptual beliefs in particular—are veridical in one’s own case but nonveridical in the BIV’s case. We will refer to these strong, stubborn intuitions about the BIV scenario as Cartesian intuitions.

As we said, the generic idea of the BIV is still commonly employed in epistemology, especially in discussions of Cartesian skepticism. In epistemological contexts, Cartesian intuitions about the BIV are routinely taken for granted, and are routinely presumed to be correct. A deep schizophrenia has arisen in contemporary philosophy, however, because meanwhile these same Cartesian intuitions have come into question in philosophy of mind.

A number of currently influential theories of mental intentionality are versions of what we will call strong externalism: roughly and generically, the view that (1) all intentionality is grounded in certain past or present causal connections between states of the cognitive system and states of the external world, and (2) there can be no mental intentionality without some suitable kind of actual connection between what is going on “in the head” and the wider environment. Strong externalist theories include (i) causal theories of content that find the necessary connection in the causal antecedents of the state, (ii) covariational theories that find the connection in certain kinds of systematic correlations between occurrences of an internal state and occurrences of an external state of affairs, (iii) teleosemantic theories that look to environmentally situated proper functions that certain internal states possess in virtue of evolutionary design, and (iv) learning-based theories that invoke internal adaptational changes in the creature’s own history. (See Stich and Warfield 1994 for a representative sample of such theories.)

Current versions of strong externalism evidently are in direct conflict with Cartesian intuitions about the BIV scenario. Advocates of various strong externalist views typically respond to this situation by “biting the bullet” and saying what their theory requires them to say about the BIV. Sometimes it is claimed, especially by advocates of causal and covariational approaches, that the BIV’s beliefs and other intentional mental states refer to various internal states or processes in the computer to which the brain has input and output connections—and to various properties of, and relations among, those states and processes in the computer (cf. Putnam, 1981). This stance denies the Cartesian intuitions in two important ways. First, it thoroughly rejects the putative match in intentionality that those intuitions say exists between your own mental life and the mental life of your BIV duplicate. Second, it asserts that the BIV’s beliefs are no less true than your own beliefs, rather than being systematically nonveridical in the way the Cartesian intuitions say they are.

Sometimes it is claimed, especially by advocates of teleosemantic and learning-based versions of strong externalism, that the kind of BIV described above—viz., one that has not come into existence as a result of evolution and natural selection—completely lacks mental intentionality. This stance obviously denies Cartesian intuitions altogether: it asserts that your BIV has no beliefs at all—let alone beliefs that match your own and are systematically nonveridical.

But Cartesian intuitions stubbornly tend to persist, despite the recent philosophical popularity of strong externalism. Our own view is that this is because Cartesian intuitions are correct, whereas strong externalism is deeply mistaken. We do not deny, however, that external factors play some important roles with respect to the some aspects of mental reference and mental intentionality: important lessons about such matters have emerged from the work of Kripke, Putnam, Burge, and other philosophers.[1] An adequate overall philosophical position about mental intentionality and mental reference, we claim, should accommodate these lessons while still thoroughly accommodating—and vindicating—Cartesian intuitions about the BIV.

In this paper we outline a position that meets these constraints, and we argue for the position largely by appeal to the way it both vindicates Cartesian intuitions and explains why they are correct. The same general position is also described and defended in Horgan and Tienson (2002), and the present paper is complementary to that one. Here we develop the position beyond what was said there, by further elaborating the treatment of narrow mental intentionality, wide mental intentionality, and mental reference. We also argue for the position in a way that complements the argumentation in Horgan and Tienson (2002); here we place much heavier emphasis on the BIV scenario and its vindication. Other philosophers who have defended positions similar in various ways to ours include McGinn (1989, 1991), Flanagan (1992), Searle (1992), Strawson (1994), Siewert (1998), Loar (forthcoming a, forthcoming b), and Pitt (forthcoming).

1. THE POSITION SKETCHED

We begin by sketching our position in broad brushstrokes, leaving various important details to be filled in later in the paper. Central in the position is the role of phenomenology or phenomenal consciousness, by which we mean those aspects of one’s mental life such that there is “something it is like” to undergo them. Briefly, the position goes as follows.

Phenomenology is narrow: it is not constitutively dependent upon anything “outside the head” (or outside the brain) of the experiencing subject. Indeed, it is not constitutively dependent upon anything outside of phenomenal consciousness itself; in this sense, it is intrinsic. Your phenomenology, being narrow and intrinsic, supervenes nomically upon physical events and processes within your brain. Hence, your phenomenology is shared in common with a BIV physical duplicate of your own brain.

Phenomenology is also richly and pervasively intentional: there is a kind of intentionality that is entirely constituted phenomenologically (we call it phenomenal intentionality), and it pervades our mental lives. Among the different aspects of phenomenal intentionality are the following. First, there is the phenomenology of perceptual experience: the enormously rich and complex what-it’s-like of being perceptually presented with a world of apparent objects, apparently instantiating a rich range of properties and relations—including one’s own apparent body, apparently interacting with other apparent objects which apparently occupy various apparent spatial relations as apparently perceived from one’s own apparent-body centered perceptual point of view. Second, there is the phenomenology of agency: the what-it’s-like of apparently voluntarily controlling one’s apparent body as it apparently moves around in, and apparently interacts with, apparent objects in its apparent environment. Third, there is conative and cognitive phenomenology: the what-it’s-like of consciously (as opposed to unconsciously) undergoing various occurrent propositional attitudes, including conative attitudes like occurrent wishes and cognitive attitudes like occurrent thoughts. There are phenomenologically discernible aspects of conative and cognitive phenomenology, notably (i) the phenomenology of attitude type and (ii) the phenomenology of content. The former is illustrated by the phenomenological difference between, for instance, occurrently hoping that Bush will fail to be re-elected and occurrently wondering whether Bush will fail to be re-elected—where the attitude-content remains the same while the attitude-type varies. The phenomenology of content is illustrated by the phenomenological difference between occurrently thinking that Bush will fail to be re-elected and occurrently thinking that Bush will succeed in being re-elected—where the attitude-type remains the same while the attitude-type varies.)

Since phenomenal intentionality is entirely constituted phenomenologically, and since phenomenology is narrow, phenomenal intentionality is narrow too. Hence, there is exact match of phenomenal intentionality between yourself and your BIV physical duplicate. This exactly matching, narrow, intentional content involves exactly matching, phenomenally constituted, narrow truth conditions. But whereas the narrow truth conditions of your own beliefs are largely satisfied, those of your BIV physical duplicate’s matching beliefs largely fail to be satisfied; thus, the BIV’s belief system is systematically nonveridical. So Cartesian intuitions about the BIV are correct.

On the other hand, exact match in narrow content between your own intentional mental states and the corresponding states in your BIV physical duplicate does not require or involve exact match in referents (if any) of all the various matching, putatively referring, thought-constituents. For instance, certain of your own occurrent thoughts that you would express linguistically using certain proper names—say, the thought that Bush is not a genius—involve singular thought-constituents whose referents (if any) are determined partly in virtue of certain external relations that obtain between you and those referents. Thus, your occurrent thought that Bush is not a genius involves a singular thought-constituent that purports to refer to a particular specific person (viz., Bush); its actually referring, and its referring to the specific individual to whom it does refer, depends upon there being certain suitable external relations linking you to a unique eligible referent (viz., Bush). A Twin-Earthly physical duplicate of yourself, in a Twin-Earthly duplicate local environment, would refer to a different individual (viz., Twin-Bush) via the corresponding singular thought-constituent of the corresponding occurrent thought. And in the case of your BIV physical duplicate, the matching singular thought-constituent fails to refer at all, because the BIV does not bear suitable externalistic relations to any suitably reference-eligible individual in its own actual environment. (Parallel remarks apply to thought-constituents that purport to refer to natural kinds, such as the thought-constituent that you yourself would express linguistically with the word ‘water’.)

For mental states involving thought-constituents for which reference depends upon externalistic factors, there are two kinds of intentionality, each involving its own truth conditions. First is the kind of intentionality already mentioned above: phenomenal intentionality, with truth conditions that are phenomenally constituted and narrow. Second is externalistic intentionality, with wide truth conditions that incorporate the actual referents (if any) of the relevant thought-constituents. Your own thought that Bush is not a genius, and the corresponding thoughts of your BIV physical duplicate and your Twin Earth physical duplicate, have matching phenomenal intentionality, with matching truth conditions. (These truth conditions are satisfied in your case and in the case of your Twin Earth duplicate, but not in the case of your BIV duplicate.) On the other hand, your own thought that Bush is not a genius and your Twin Earth duplicate’s corresponding thought do not have matching externalistic intentionality, because the externalistic truth conditions of these respective thoughts do not match: the truth value of your own thought depends upon the intelligence level of Bush, whereas the truth value your Twin Earth duplicate’s corresponding thought depends upon the intelligence level of an entirely different individual, viz., Twin-Bush. (Each thought’s wide truth conditions are indeed satisfied.) As for your BIV duplicate’s thought, it lacks externalistic intentionality and wide truth conditions, because its singular thought-constituent purporting to refer to a person called ‘Bush’ does not actually refer at all.

If the picture we have sketched can be satisfactorily elaborated, then the resulting position will meet the two constraints we mentioned in the introduction. On one hand, the position will vindicate Cartesian intuitions: you and your BIV physical duplicate have exactly matching conscious occurrent mental states with exactly matching narrow intentional content, and for vastly many of the BIV’s conscious occurrent beliefs, the narrow truth conditions are not satisfied. On the other hand, the position also will accommodate externalistic aspects that govern the reference of certain thought-constituents, such as thought-constituents that purport to refer to individuals or to natural kinds. Your own conscious occurrent mental states and the corresponding ones of your BIV duplicate do not have matching externalistic intentionality. On the contrary, the BIV’s mental states lack externalistic intentionality altogether.

2. METHODOLOGICAL REMARKS

Before we elaborate the position just sketched, let us comment briefly about the structure of our argument for this position, and about how this argument fits into the wider dialectical landscape of philosophical views and arguments concerning mental intentionality. To begin with, we believe that a strong case can be made for phenomenal intentionality—and for its narrowness, its richness, and its pervasiveness in human mental life—without reliance on Cartesian intuitions about the BIV. One important line of argumentation involves focusing the reader’s introspective attention on certain specific actual or potential experiences, as a way of prompting appreciation that such experiences reveal the presence of one or another specific kind of phenomenal intentionality—in some cases involving forms of non-sensory phenomenology that are more subtle, less starkly vivid, and thus more easily overlooked than is sensory phenomenology.

For instance, Galen Strawson (1994) discusses a form of non-sensory, inherently intentional, phenomenology that he calls “understanding experience.” He argues that understanding and other related kinds of occurrent mental states and processes are very commonly, if not always, laden with distinctive phenomenal character that is non-sensory and inherently involves the intentional content of what is understood. He points out, for example, the phenomenological difference between hearing speech in a language that one does not understand and hearing speech in a language that one does understand. Imagine two people side by side hearing the same spoken sequence of sounds, with one of them understanding the language and the other one not. At a certain relatively raw sensory level, their auditory experience is phenomenologically the same; the sounds are the same, and in some cases may be experienced in much the same way qua sounds. Yet it is obvious introspectively that there is something phenomenologically very different about what it is like for each of them: one person is having understanding experience with the distinctive phenomenology of understanding the sentence to mean just what it does, and the other is not.

Such examples are readily multiplied. (See, for instance, the works cited above at the end of the introductory section.) But even though there is already a strong case to be made for phenomenal intentionality without appeal to the BIV, nonetheless the fact that Cartesian intuitions about the BIV are so common and so persistent is itself dialectically important; it provides additional, reinforcing, evidence. Here is why. The persistence and widespread occurrence of Cartesian intuitions is a datum that calls out for explanation. The initial, albeit defeasible, presumption about these intuitions is that their persistence and their ubiquity reflect their correctness—although it remains possible that in the end, the best explanation will be a “debunking” explanation that treats them as erroneous. So if we provisionally assume that the Cartesian intuitions are correct, and we then seek out a plausible non-debunking explanation for their occurrence, by far the most plausible source for such an explanation is phenomenology. The relevant explanatory hypothesis asserts, about somone with Cartesian intuitions, (i) that one is judging that a BIV that is a physical duplicate of oneself would also be a phenomenal duplicate of oneself, and (ii) that one’s Cartesian intuitions reflect one’s visceral appreciation that a phenomenal duplicate of oneself would, by virtue of having a mental life phenomenologically exactly like one’s own, thereby have a mental life that exactly matches one’s own intentionally. (The visceral appreciation mentioned in (ii) need not necessarily rise to the level of explicit belief.) This explanatory hypothesis provides the most plausible—indeed, really the only plausible—non-debunking explanation of the Cartesian intuitions. Thus, the persistence and widespread occurrence of these intuitions provides significant evidential support for the explanatory hypothesis itself, via “inference to the only plausible non-debunking explanation.”

Of course, the ultimate strength of this evidential support depends upon whether or not the position sketched in section 1 can be developed in a coherent and credible way—in some way that, inter alia, smoothly and naturally accounts for the role of externalistic factors vis-à-vis mental reference and mental intentionality. Addressing this task will be the business of the next section. But before turning to that, we have two final dialectical observations.

First, we realize that if there were a good case in favor of strong externalism about mental intentionality, such a case would count against our own position—and would also constitute grounds for believing that the best explanation of lingering Cartesian intuitions about the BIV must be some kind of debunking explanation. But our own view is that strong externalism is seriously lacking in solid epistemic support—its widespread recent popularity notwithstanding. Although the kinds of considerations often cited as grounds for strong externalism do indeed show that external factors are constitutively involved in the fixation of reference for some reference-purporting thought-constituents, we believe that such considerations fall far short of warranting strong externalism itself. The discussion to follow in section 3 should make clear why this is so.

Second, it bears emphasis that even among those philosophers who maintain (as we do) that there is a narrow kind of mental intentionality that is wholly constituted by phenomenology, there are important theoretical differences. To some extent, these differences are reflected in the extent to which the various alternative approaches to phenomenal intentionality are able to accommodate Cartesian intuitions about the BIV. We will return to this matter briefly at the end of the paper.

3. THE POSITION ELABORATED

We will now develop further the position sketched above. In doing so we will use Cartesian intuitions about the BIV as a guide and motivator: we will elaborate the position in ways that seem needed in order to accommodate Cartesian intuitions—and that also seem independently plausible, both phenomenologically and theoretically. What we say here will develop the initial sketch into a somewhat more complete picture, but will still leave many important matters open for yet further exploration.

It will be useful to articulate the position by means of a device briefly employed already in section 1, viz., by discussing two different kinds of physically possible physical duplicates of yourself: not only a BIV physical duplicate (of the sort described in the second paragraph of this paper), but also a Twin Earth physical duplicate.[2]

We assume, here and below, what was said in section 1; our present purpose is to elaborate that discussion without repeating it. In particular, we assume that a physical duplicate of your own brain, undergoing exactly the same physical inputs and exactly the same internal physical processes, would also be a phenomenal duplicate: it would undergo phenomenology that exactly matches your own.

3.1. Matching Phenomenal Intentionality: Self-Indexical Aspects and Phenomenally Constituted Attribute-Reference

In considering the extent of mental match between yourself and your BIV phenomenal duplicate, it is useful to begin by focusing on matching perceptual experience, and on matching beliefs whose intentional contents are the contents of one’s current perceptual experience. Phenomenologically, perceptual experience is richly presentational: it presents, to the experiencing subject, a richly textured apparent world of apparent concrete objects apparently instantiating numerous properties and relations. For instance, experience presents various apparent objects apparently instantiating various shape-properties, size-properties, relative-position relations, and relative-orientation relations. Among the experientially presented apparent objects is one’s own apparent body, with its various apparent component parts—a body that is apparently under one’s voluntary control. Some of the relations apparently instantiated by the various apparent objects—for instance, relative-position relations and relative-orientation relations—are experientially presented within a self-oriented reference frame. Thus, apparent objects are experientially presented as lying at various distances from oneself (as well as from one another, as presented from one’s own visual point of view), and as apparently oriented various ways vis-à-vis oneself (as well as vis-à-vis one another, as presented from one’s own visual point of view). This point of view is what Husserl called the “zero-point” in this self-oriented reference frame. Up and down in this reference frame normally depend heavily upon kinesthetic/tactile aspects of phenomenology, in combination with visual aspects: roughly, down is the direction that one’s own apparent body apparently tends to move of its own accord, and the direction of a surface to which one’s apparent body apparently tends to stay attached when apparently not moving relative to that surface.

Two points about such experientially presented properties and relations are of key importance. First, often their presentation in experience includes a self-indexical aspect, involving the self-oriented nature of the experiencing subject’s perceptual reference frame. Because of this indexical element, if two different experiencing subjects have exactly similar phenomenology, the relevant kind of intentional match between them will not be literal identity of intentional content. Rather, two distinct selves are involved, and hence two distinct self-oriented experiential reference-frames. The intentional match between them consists in the fact that the corresponding intentional contents of the two distinct subjects’ corresponding perceptual-experiential states are, so to speak, “identical apart from differences in the identity of the experiencer who is the self in the self-oriented perceptual reference frame.” This theme surely deserves further exploration and elucidation, but that is a task we will not pursue here.[3]

The second key point is this. When experience presents various apparent objects as apparently instantiating properties and relations such as shape-properties and relative-position relations, experience thereby acquaints the experiencing subject with such properties and relations, and this mental acquaintance-relation grounds mental reference to these properties and relations. Such mental reference is wholly constituted phenomenologically.[4] It makes no difference, so far as this phenomenally constituted and reference-grounding form of acquaintance is concerned, whether or not the relevant experiential presentations are veridical. In the case of your BIV phenomenal duplicate, for instance, the perceptual-experiential presentations are radically illusory: there are no real objects that are really perceived by that experiencing subject and that really instantiate the relevant properties and relations. But no matter: your BIV’s duplicate’s perceptual experience acquaints the BIV with shape-properties and relative-position properties just as much as your own perceptual experience does, even though this acquaintance occurs via radically nonveridical experiences of merely apparent instantiations of these properties and relations by merely apparent objects. And for the BIV, such experiential acquaintance with the properties and relations grounds mental reference to them—just as it does for you. Experientially presented apparent instantiation of the properties and relations suffices to acquaint the experiencing subject with them, and thus suffices to ground mental reference to them, whether or not the experiencing subject is ever experientially presented with actual instantiations of them.[5]

In addition to phenomenally constituted reference to certain properties and relations that is based on direct experiential acquaintance, there are further kinds of phenomenally constituted reference to properties and relations too. The experiencing subject conceives these properties and relations by way of aspects that go beyond what is directly experientially presented—for instance, dispositional aspects and causal-role aspects. But the conceiving of such properties and relations, with such aspects, is itself phenomenally constituted (and hence narrow), and largely rests on the experiencer’s capacity for “conceptual bootstrapping” on the basis of phenomenally constituted mental reference to those properties and relations (including causation) that are most directly presented in experience.

It is an important philosophical question which kinds of properties and relations are ones to which creatures with human-like phenomenology bear phenomenally constituted, reference-grounding, experiential-acquaintance relations. Although we cannot pursue this large topic at any length here, we take it that the range of such properties and relations is very extensive. It appears to include, inter alia, temporal relations, causal relations, properties like being a temporally persisting object, being an animal, being an agent, and being a person, numerous artifactual kinds like being a container and being a table, and numerous social relations and properties like being friend of, being a boss of, and being a politician.[6] It also appears to include numerous language-involving properties and relations, such as uttering a meaningful statement and speaking a language I understand.

Many of these properties and relations involve aspects that are not fully and directly presented in experience (as mentioned two paragraphs ago). For instance, being a container is partly a matter of dispositional capacities to contain something, and paradigmatically (if not inevitably) is partly a matter of something’s having its causal origin as an artifact that was deliberately created in order to contain things. Although such properties and relations typically are not presented in perceptual experience as immediately and straightforwardly as are features like position and shape, this certainly does not mean that they figure in phenomenology in a “merely theoretical” or inferential way, as opposed to an “observational” way. On the contrary, the presentational aspects of perceptual experience are rife with potentialities. You experience apparent enduring objects as having back sides, even though those sides are not directly presented; you experience an apparent cup as a cup, even though its being a cup includes certain dispositional features and causal-origin features that are not directly experientially presented; you experience other apparent bodies that suitably resemble your own apparent body as persons who are acting for intelligible reasons, even though many key features that make for personhood are not directly experientially presented, and even though the presumptive mental lives of others are not directly experientially accessible to you; and so on.

Because of the extensive range of properties and relations to which the experiencing subject bears the relation of phenomenally constituted mental reference, there are numerous potential thoughts that have only narrow, phenomenal, intentionality. (Thoughts that have both phenomenal intentionality and externalistic intentionality will be discussed below.) Roughly, these are thoughts that are expressible linguistically using only (i) logical vocabulary, (ii) predicates expressing properties and relations to which the experiencer can mentally refer in a phenomenally constituted way, and (iii) certain first-person indexical expressions. So for instance, you might have a thought you could express linguistically by saying “A picture is hanging crooked on a wall directly in front of me.” Your BIV phenomenal duplicate and your Twin Earth phenomenal duplicate would have corresponding occurrent thoughts with exactly matching, phenomenally constituted (and hence narrowly constituted), truth conditions. In terms of logical form, these matching thoughts have matching, doubly existential, contents involving the respective experiencing subjects as the respective referents of the first-person indexical thought-constituent: there is an x and there is a y such that x is a picture, y is a wall directly in front of me, and x is hanging crooked on y (relative to the up-down axis of my self-oriented visual/kinesthetic reference frame).[7]

3.2. Mental Reference to Concrete Particulars: Grounding Presuppositions and Externalistic Factors.

Although some kinds of mental reference are fully constituted by phenomenological factors alone (and hence purely narrowly), other kinds are constituted in a way that involves not only phenomenology but also certain externalistic factors. Singular mental reference to concrete particulars generally (perhaps always) works this way. Suppose, for example, that you have an occurrent thought that you could express linguistically by saying “That picture is hanging crooked,” where the singular thought-constituent expressible linguistically by ‘that picture’ purports to refer to a picture on the wall directly in front of you. This thought-content involves certain phenomenally constituted presuppositions, which we call grounding presuppositions, that must be satisfied in order for the singular thought-constituent to refer: roughly, there must be an object at a certain location relative to oneself (a location that one could designate linguistically by a specific use of the place-indexical ‘there’), this object must be a picture, and there must not be any other picture at that location that is an equally eligible potential referent of ‘that picture’. If these grounding presuppositions are satisfied by some specific concrete particular in your ambient environment—some particular object that is a picture and is uniquely suitably located—then your singular thought-thought constituent thereby refers to that very object. Which object your thought-constituent refers to, if any, thus depends jointly upon two factors, one phenomenally constituted and one externalistic: on one hand, the phenomenally constituted grounding presuppositions, and on the other hand, the unique actual object in your ambient environment that satisfies those presuppositions.

Compare the corresponding singular thought of your Twin Earth phenomenal duplicate. The corresponding singular thought-constituent also refers to something, because on Twin Earth too there is a unique object that satisfies the matching, phenomenally constituted, grounding presuppositions. However, that thought-constituent refers to a different concrete particular than yours does, viz., the specific picture that is suitably located in front of your Twin Earth phenomenal duplicate rather than in front of yourself. The referents are different, for you and for your Twin Earth duplicate, because in the respective cases two different objects respectively satisfy the matching, self-indexical-involving, grounding presuppositions. Relative to you, the presuppositions are satisfied by the picture in front of you; relative to your duplicate, they are satisfied by the picture in front of her/him.

Next, compare your BIV phenomenal duplicate’s singular thought corresponding to your own singular thought, and the BIV’s singular thought-constituent corresponding to your own thought-constituent that you could express linguistically by ‘that picture’. For the BIV duplicate too, this thought-constituent has phenomenally constituted grounding presuppositions that match those of yours. However, nothing in the BIV’s ambient environment satisfies those presuppositions. Hence, in the case of your BIV phenomenal duplicate, that thought-constituent fails to refer to anything.

According to the conception of mental intentionality we are here setting forth, these observations are pervasively generalizable. In general, singular thought-constituents have phenomenally constituted grounding presuppositions, and the referent (if any) of such a thought-constituent is the unique object (if any) in the experiencer’s own environment that satisfies the applicable, phenomenally constituted and self-indexical involving, grounding presuppositions. Needless to say, the nature of grounding presuppositions is an important philosophical topic worthy of further investigation. We will return to this topic briefly in section 3.6.

3.3. Mental Reference to Natural Kinds.

Section 3.2 focused on singular thought-constituents that purport to refer to concrete individuals. These thought-constituents normally would be expressed in language via singular terms, such as names or definite descriptions. Let us return now to predicative thought-constituents, those that purport to refer to properties, relations, or kinds. These include both general thought-constituents (roughly, ones that normally would be expressed in natural language via general terms such as verbs or adjectives, and in formal language as predicates) and certain singular thought-constituents that purport to refer to kinds, properties, or relations (thought-constituents expressible in natural language, for instance, by certain mass terms like ‘water’).

Although many predicative thought-constituents refer in a way that is wholly constituted phenomenally, not all of them do. Among those that do not, as is widely acknowledged in light of the writings of Kripke and Putnam, are certain thought-constituents that purport to refer to natural kinds—for instance, thought-constituents expressible in language by terms like ‘water’ or ‘gold’. The key points in the preceding section carry over to these kinds of predicative thought-constituents, mutatis mutandis. Here too, the relevant thought-constituents have certain phenomenally constituted, reference-fixing, grounding presuppositions; and such a thought-constituent refers to the unique kind, property, or relation (if any) that satisfies those presuppositions. Here too, it is an externalistic matter whether or not the presuppositions happen to be satisfied at all; and it is an externalistic matter what (if anything) satisfies them. Here on Earth, the phenomenally constituted grounding presuppositions governing the thought-constituent expressible linguistically by the word ‘water’ are satisfied by H₂0. On Putnam’s Twin Earth, the matching phenomenally constituted grounding presuppositions are instead satisfied by a different natural kind: XYZ. So for human experiencers on Earth, the relevant thought-constituent refers to H₂O, whereas for twin-human experiencers on Twin Earth, the corresponding thought-constituent refers instead to XYZ. And for your BIV phenomenal duplicate, the corresponding thought-constituent does not refer at all, because nothing satisfies those grounding presuppositions.

Delineating the range of predicative thought-constituents whose reference depends in part upon externalistic connections between the experiencer and the experiencer’s environment, and also the range of predicative thought-constituents whose reference is instead constituted in a wholly phenomenal way, is an important philosophical task. Some cases are especially vexing, and various competing positions are possible. For instance, do predicative thought-constituents that purport to refer to colors refer in a phenomenally constituted way, or instead in an externalistic way? Or do they, perhaps, fail to refer to any genuine properties at all?[8]

The approach to mental intentionality we are describing here is officially neutral about predicative thought-constituents that purport to refer to colors and to other so-called “secondary qualities.” But again, two key tenets of the approach are these. First, regardless of the exact range of predicative thought-constituents whose reference is partly externalistic, there are vastly many predicative thought-constituents whose reference is instead constituted wholly phenomenally. Second, when the reference of a predicative thought-constituent is partly externalistic, this is a matter of there being some unique kind, property, or relation in (or instantiated in) the experiencer’s ambient environment that satisfies certain phenomenally constituted, reference-fixing, grounding presuppositions.

3.4. Mental Reference and Social Deference

Perceptual experience presents to the experiencer an apparent world that is richly social: a world full of apparent persons apparently interacting with each other and with the experiencer in multifarious ways. Experience presents many of these apparent persons as apparently interacting with oneself linguistically; one experiences oneself as belonging to a linguistic community. Occasionally one experiences oneself as apparently employing language in a deferential way—as employing terminology that purports to refer to individuals, kinds, properties, or relations that one does not know much about oneself but that one believes are known about, and identifiable by, others in one’s apparent linguistic community who have suitable expertise. Thus, the apparent use of language in an apparently socially-deferential way, within an apparent linguistic community that apparently engages in the “division of linguistic labor,” are all aspects of the experiencing subject’s overall phenomenology. So these aspects are all present in your BIV duplicate’s experience as well as in your own—even though, in the case of the BIV, such experience is systematically nonveridical.

How do phenomenally constituted grounding presuppositions work, in the case of apparently language-mediated constitutents of thought that are employed in an apparently socially deferential way? Suppose, for instance, that you have an occurrent thought that you could express linguistically by saying “Elm trees grow around here,” and that you have no ability yourself to distinguish elm trees from various other kinds of trees. Consider your Twin Earth phenomenal duplicate’s matching thought, and your BIV phenomenal duplicate’s matching thought, each with a matching thought-constituent corresponding to your own ‘elm tree’ thought-constituent. How do the phenomenally constituted grounding presuppositions work, ones that you yourself share in common these phenomenal duplicates?

Three interrelated factors come into play here. First are phenomenally constituted grounding presuppositions governing the thought-constituent you would express with the word ‘tree’, and also governing the corresponding thought-constituent of your Twin Earth phenomenal duplicate and your BIV phenomenal duplicate. These presuppositions already put substantial constraints on what could count as a referent of the thought-constituent purporting to refer to elm trees. (Dogs or automobiles, for instance, couldn’t be elm trees.) Likewise, mutatis mutandis, for the corresponding thought-constituent of your Twin Earth phenomenal duplicate and your BIV phenomenal duplicate.

Second are phenomenally constituted grounding presuppositions about the existence of actual experts in one’s ambient environment, about these experts belonging to a genuine linguistic community that includes you too, and about their using ‘elm tree’ in some specific, referentially determinate, way. When these presuppositions are satisfied, your own thought-constituent that purports to refer to elm trees has its referents determinately fixed by the actual usage of these actual experts in your environment. This happens for you, and (with different referents) for your Twin Earth phenomenal duplicate. But it does not happen for your BIV phenomenal duplicate, because for it those grounding presuppositions are not satisfied; it is not really a member of a genuine linguistic community at all.

Third are certain counterfactual facts about your phenomenology: facts about experiences that you would have if you were to have various kinds of investigation-experiences and/or expert-consultation experiences with respect to what counts as an elm tree. Such facts too reflect specific constraints on what counts as an elm tree, insofar as they provide evidence about how experts use the term ‘elm tree’ that is not at odds with how experts actually use that term. So in effect, these counterfactual-phenomenal facts underwrite assumptions that defeasibly have the status of grounding presuppositions: assumptions that have this status provided that they are not in conflict with the actual usage of actual experts in the experiencer’s actual environment. For you yourself and for your Twin Earth phenomenal duplicate, these defeasibility conditions are indeed met, because actual expert usage conforms to what the relevant counterfactual phenomenology would present such usage to be. So, counterfactual phenomenology reflects specific, linguistically deferential, constraints on the reference-class of your own ‘elm-tree’ thought-constituent and on the reference-class of your Twin Earth duplicate’s corresponding thought-constituent.

What about your BIV phenomenal duplicate? It here becomes important that you yourself and your BIV duplicate are exactly similar not only with respect to your actual ongoing physical brain-processes and phenomenology, but also are robustly counterfactually similar. (See the second paragraph of the present paper.) Thus, the following is true of both you yourself and your BIV duplicate: if you (i.e., either one of you) were to have the experience of apparently consulting an expert about the apparent objects you apparently call “trees,” then you would have the experience of apparently being shown or told some key distinguishing features of the apparent objects that you experience the apparent experts as apparently calling “elm trees.”

Now, in the case of your BIV phenomenal duplicate, there is not a discrepancy between (i) such counterfactual-phenomenology facts and (ii) actual linguistic practice by actual experts in the BIV phenomenal duplicate’s actual linguistic community. There is no such discrepancy because there does not exist a real linguistic community of which your BIV phenomenal duplicate belongs. So the defeasibility conditions that govern the counterfactual-phenomenology based, defeasibly grounding, presuppositions are not violated in the case of the BIV; thus, those assumptions do indeed have the status of grounding presuppositions. Counterfactual-phenomenological facts about the BIV thereby provide additional grounding presuppositions governing the BIV’s ‘elm tree’ thought-constituent, over and above those provided by facts about the BIV’s actual phenomenology. These further presuppositions involve the ways that apparent experts in the BIV’s apparent linguistic community apparently would employ the apparent words ‘elm tree’. The BIV’s counterfactual phenomenology—the phenomenology that would ensue were the BIV to apparently seek out information about expert linguistic usage—places real constraints on what is required in order for something to be a referent of the BIV’s ‘elm tree’ thought-constituent; it does so even though the apparent linguistic community is not itself real. The key point is that the operative, phenomenally constituted, grounding presuppositions rest upon more than the BIV’s actual experience; they also depend partly upon certain counterfactual aspects of its experience.[9]

3.5 Mental Reference to Theoretical Entities

Apparent ongoing scientific inquiry, together with the apparently linguistically mediated dissemination of the apparent results of such apparent inquiry, are aspects of the apparent world that is experientially presented to the experiencing subject. As a result of such experiences—including experiences of apparently studying science at a university, of apparently reading about esoteric theories with names like “Loop Quantum Gravity” in an apparent magazine apparently called Scientific American, of apparently conversing with apparent persons who apparently speak of themselves as “physicists,” and the like—your BIV duplicate has a rich stock of thought constituents that purport to refer to various quite esoteric objects, processes, properties, and relations, many of which are not experienced observationally. Putative reference to such esoteric entities does not require any actual, externalistic, reference-constituting connections to them, any more than such externalistic connections are required for putative reference to tables, chairs, or other persons. Instead it arises via the experiencer’s capacity for conceptual bootstrapping. Such bootstrapping begins with phenomenally constituted grounding presuppositions governing thought-constituents purporting to refer to concrete objects and natural kinds that experience presents as directly observable. From there the bootstrapping proceeds up—often via apparent social-linguistic mediation, and often in a way that rests in part upon the fact that the experiencer can refer mentally to the relation of causation purely by virtue of experiential presentations of apparent instantiations of it—to phenomenally constituted grounding presuppositions governing thought-constituents that purport to refer to various kinds of non-observable “theoretical” entities and properties.

Your BIV phenomenal duplicate has a full range of phenomenally constituted occurrent beliefs that exactly match your own, including scientific beliefs. For you and your BIV phenomenal duplicate, these matching beliefs have exactly matching, phenomenally constituted, grounding presuppositions.

3.6. Phenomenal Intentionality, Externalistic Intentionality, and Two Kinds of Truth Conditions

As we said in section 3.1, some potential thoughts have only the narrow, phenomenally constituted, kind of intentionality. Roughly, these are thoughts none of whose reference-purporting constituents are such that their reference (if any) depends in part upon externalistic factors. But in addition, there are numerous potential thoughts employing reference-purporting constituents whose reference (if any) does partially depend upon such factors. (We will call these externalistic reference-purporting thought-constituents.) Such thoughts are capable of two kinds of intentionality: phenomenal intentionality and externalistic intentionality.[10] Phenomenal intentionality is narrow: there is a match in phenomenal intentionality between your own occurrent thought (say, a thought you could express linguistically by saying “That picture is hanging crooked”) and the corresponding occurrent thoughts of your Twin Earth phenomenal duplicate and your BIV phenomenal duplicate. Externalistic intentionality, on the other hand, is wide: it incorporates the specific referents (if any) of thought-constituents whose reference is fixed in a way that involves external factors. Your picture-thought and your Twin Earth duplicate’s corresponding picture-thought are about different pictures, and thus these corresponding thoughts have non-matching externalistic intentionality. Your BIV phenomenal duplicate’s corresponding picture-thought fails to refer to any picture at all, and hence does not have externalistic intentionality.

Thoughts with externalistic reference-purporting constituents are capable of two kinds of truth conditions, corresponding to the two kinds of intentionality. Phenomenal intentionality involves narrow truth conditions. These have “built into” them all the phenomenally constituted grounding presuppositions governing the given thought’s externalistally reference-purporting thought-constituents. Insofar as such narrow truth conditions are formulable linguistically (whereof more below), the formulation will employ only these kinds of vocabulary: (i) logical expressions, (ii) predicative expressions designating properties and relations to which the experiencer can mentally refer non-externalistically, and (iii) certain first-person indexical expressions. Take, for instance, a thought that you could express in language by saying “That picture is hanging crooked” where ‘that picture’ purports to refer to a picture directly in front of you. The narrow truth conditions for this thought would be expressible something like this: “There is a unique object x, located directly in front of me and visible by me, such that x is a picture and x is hanging crooked (relative to my visual/kinesthetic up/down axis).” Your own picture-thought, your Twin Earth phenomenal duplicate’s picture-thought, and your BIV duplicate’s picture-thought all match with respect to these truth conditions; i.e., the truth conditions are just the same, apart from the different referents of the first-person indexical.

Externalistic intentionality, on the other hand, involves wide truth conditions. The wide truth conditions of a given thought incorporate the specific satisfiers (if any) of the phenomenally constituted grounding presuppositions governing the externalistic reference-purporting constituents of that thought. In order for the thought’s wide truth conditions to be met, those specific satisfiers must play a suitable truth-making role. Thus, your picture-thought and your Twin Earth phenomenal duplicate’s corresponding picture-thought have non-matching wide truth conditions, because they respectively involve different pictures—in one case, the picture in front of you, and in the other case, the picture in front of your Twin Earth duplicate. As for your BIV phenomenal duplicate’s picture-thought, it lacks wide truth conditions altogether (and hence lacks externalistic intentionality), because the pertinent grounding presuppositions are not satisfied. (Likewise, mutatis mutandis, for your own thought that you could express linguistically by saying “Water is good to drink,” and the corresponding thoughts of your Twin Earth phenomenal duplicate and your BIV phenomenal duplicate.) For a thought that you could express in a sentence of your own language, normally the thought’s wide truth conditions can be formulated homophonically, via that very sentence—because normally a word or phrase expressing an externalistically referential thought-constituent will be externalistically referential itself.[11]

Is it in general possible to provide compact, cognitively surveyable, formulations of the narrow truth conditions for thoughts and for other intentional mental states? We doubt it, and we deny that this needs to be possible. Let us take up these points in turn: why we doubt it, and why the lack of such formulations would not be a problem. Here are two reasons to doubt that narrow truth conditions can always be tractably, compactly, formulated. First, ordinary perceptual experience is normally so phenomenally-intentionally rich that its intentional content defies complete description. The phenomenally constituted grounding presuppositions constituted by perceptual experience inherit this feature, and often one’s ongoing intentional states and processes rely heavily upon enormous portions of this presuppositional background. (Think, for instance, of the presuppositional background at work in your mental life as you apparently hike along an apparent mountain trail—apparently constantly positioning your feet in optimal ways given the apparent local terrain both near your apparent feet and further up the apparent trail, apparently pausing periodically to take in the apparent view and examine various specific apparent objects, etc.) Second, many of the putatively referential thought-constituents deployed in thought, especially those thought-constituents involving apparently social aspects of the apparent world presented in experience, are heavily interconnected with other putatively referential constituents of other potential thoughts, in a holistically ramifying web of interlocking grounding presuppositions. So the presuppositional background of a given thought is often enormous, involving a significant portion of the web. (Think, for instance, of the vast presuppositional background lying being the thought you could express linguistically by saying “Even some Republicans in the U.S. Congress are apt to balk at Bush’s plan to eliminate federal taxes on investement dividends, because such a plan would be likely to harm the U.S. economy more than help it.”)

What exactly are narrow truth conditions, if they typically are not compactly formulable linguistically in a way that can be neatly plugged into the right side of statements of the form “Statement ‘S’ is true iff…”? Well, the truth conditions for a thought are the various possible ways the world might be each of which would accord with the thought’s intentional content. As long as there is a reasonably determinate range of ways the world might be that count as potential truth-makers for the thought, then the possibilities within this range collectively constitute the thought’s truth conditions. (We say reasonably determinate because intentional content quite often exhibits some degree of vagueness.[12]) Truth conditions, thus understood, certainly need not be expressible in compact linguistic formulations; indeed, they need not be fully expressible even in infinitary linguistic formulations.[13]

If, as we are suggesting, the background presuppositions figuring in the narrow content of intentional mental states typically cannot be spelled out in any tractable way, and if indeed these presuppositions typically are too complex and too ramified to be cognitively surveyable, do these facts prevent humans (and BIV duplicates of human brains) from possessing and competently deploying thought-constituents governed by such background presuppositions? Not at all. Rather, what’s required for cognitive/conceptual competence is that the cognitive agent make judgments, and possess judgment-forming dispositions, that accord reasonably accurately with the narrow truth conditions of the judgments—modulo available evidence. Humans meet this standard, and so would BIV duplicates of human brains.[14] The BIV duplicates’ judgments would very often be mistaken, of course, but nonetheless would be the right judgments to make given the available experiential evidence.

Because thoughts with externalistic reference-purporting constituents are capable of two kinds of truth conditions—narrow and wide—they also are capable of two kinds of truth or falsity. First, such a thought is narrowly true just in case there exist entities that (i) satisfy the operative grounding presuppositions that govern the thought’s externalistic reference-purporting constituents, and (ii) have the features that the thought affirms to them. It is narrowly false just case it is not narrowly true. Second, if (and only if) there do exist entities satisfying the operative grounding presuppositions, the thought has wide truth conditions too.[15] So if the grounding presuppositions are satisfied, then the thought is widely true just in case the referents of its externalistically referring thought-constituents have the features that the thought attributes to them, and it is widely false just in case those referents do not have the features that the thought attributes to them. (If the thought’s grounding presuppositions are not satisfied, then it does not have externalistic truth conditions; so it is neither widely true nor widely false.[16])

When grounding presuppositions are satisfied, narrow truth and falsity normally run smoothly in tandem with wide truth and falsity: a thought with externalistic reference-purporting constituents is narrowly true just in case it is widely true, and is narrowly false just in case it is widely false. Consider, for instance, the thought that you could express linguistically by saying “Water is good to drink.” This thought is narrowly true just in case (i) there exists a kind of stuff uniquely satisfying the grounding presuppositions governing your ‘water’ thought-constituent, and (ii) this stuff is good to drink. And the thought is widely true just in case that very stuff (viz., H₂O) is good to drink.[17]

On the other hand, when the grounding presuppositions are not satisfied for a thought with externalistic reference-purporting constituents, then the thought is narrowly false and is neither widely true nor widely false. It is narrowly false because the grounding presuppositions are built directly into the narrow truth conditions themselves, and fail to be met. And it is neither widely true nor widely false because it lacks wide truth conditions (and wide falsity conditions). This is precisely the situation, on a massive scale, with respect to your BIV phenomenal duplicate’s thoughts that have externalistic reference-purporting thought-constituents. Those thoughts all are narrowly false; and because they lack wide truth conditions, they all are neither widely true nor widely false.

4. CONCLUSION

The position we have outlined meets the desiderata we set forth at the outset: it accommodates and vindicates Cartesian intuitions about the BIV, it accommodates the externalistic factors that play a constitutive role in the mental reference of many thought-constituents, and it is independently plausible both phenomenologically and theoretically. The key to this package is the recognition that thoughts with externalistic reference-purporting constituents normally have two kinds of intentionality, each with its own truth conditions: phenomenal intentionality and externalistic intentionality. Phenomenal intentionality is narrow, being wholly constituted by phenomenology alone. Your BIV duplicate has phenomenal intentionality that matches your own—with matching, phenomenally constituted, truth conditions. Externalistic intentionality on the other hand, is wide: its truth conditions incorporate the specific satisfiers, if any, of the phenomenally constituted grounding presuppositions that govern externalistic reference-purporting thought-constituents. Your BIV duplicate lacks externalistic intentionality, because the grounding presuppositions governing its thoughts are not satisfied.

On our account, the BIV’s thoughts with externalistic reference-purporting constituents are narrowly false, and are neither widely true nor widely false. This captures well the radical kind of nonveridicality that intuition attributes to the BIV’s external-world beliefs. Their falsity is not the ordinary kind, in which a belief’s grounding presuppositions are satisfied but its satisfiers do not have the features that the belief attributes to them. (In ordinary cases, a false belief is both narrowly and widely false.) Rather, it is a deeper kind of falsity, in which the grounding presuppositions themselves fail to be satisfied.

Our account rests heavily and essentially upon the contention that mental reference to many properties and relations—including various spatiotemporal-location properties, shape-properties, size-properties, artifact-properties, and personhood-involving properties—is wholly constituted by phenomenology alone. Even systematically nonveridical phenomenology, as in the case of the BIV, provides reference-constituting acquaintance with such properties and relations. These properties and relations, in turn, are the ones figuring in the phenomenally constituted grounding presuppositions at work in narrow, phenomenal, intentionality.

Some philosophers recently have put forth views that resemble ours in positing some form of phenomenal intentionality, but that differ from ours by not honoring the Cartesian intuition that the BIV has beliefs that are systematically nonveridical. These views depart from our own position either (i) by treating all mental reference externalistically (e.g., Loar, forthcoming a, forthcoming b), or else (ii) by narrowing the extent of phenomenally constituted mental reference in such a way that the BIV’s thoughts allegedly refer to what goes on in the computer that generates its inputs, and hence are actually veridical (e.g., Chalmers, forthcoming). Addressing these alternative approaches to phenomenal intentionality is a task for another occasion, but meanwhile the following point deserves emphasis: our own approach has the significant advantage, over these others, of thoroughly accommodating Cartesian intuitions.

Versions of strong externalism about mental reference and mental intentionality have been very influential in recent philosophy, and strong externalism poses a challenge to Cartesian intuitions about the BIV. We have argued here that this challenge can be met, in a way that is independently plausible, vindicates Cartesian intuitions, and incorporates the reference-fixing roles often played by externalistic factors. If the position described here is correct, then strong externalism itself is seriously in error. We maintain that recent insights about externalist aspects of reference should be disentangled from strong externalism, and should be tethered instead to the quite different philosophical picture we have outlined. The most fundamental kind of mental intentionality is narrow, phenomenal, intentionality.[18]

Terence Horgan, University of Arizona

John Tienson, University of Memphis

George Graham, Wake Forest University

References

Chalmers, D. (1996). The Conscious Mind, Oxford University Press.

Chalmers, D. (2002). “The Components of Content,” in D. Chalmers, ed., Philosophy of Mind: Classical and Contemporary Readings, Oxford University Press.

Chalmers, D. (forthcoming). “The Brain in a Vat as a Metaphysical Hypothesis.”

Davies, M. and Humberstone, I. (1980). “Two Notions of Necessity,” Philosophical Studies 38, 1-30.

Flanagan, O. (1992). Consciousness Reconsidered, MIT Press.

Horgan, T. and Tienson, J. (1996). Connectionism and the Philosophy of Psychology, MIT Press.

Horgan, T. and Tienson, J. (2002). “The Intentionality of Phenomenology and the Phenomenology of Intentionality,” in D. Chalmers, ed., Philosophy of Mind: Classical and Contemporary Readings, Oxford University Press.

Jackson, F. (1998). From Metaphysics to Ethics: A Defense of Conceptual Analysis, Oxford University Press.

Loar, B. (forthcoming a). “Phenomenal Intentionality as the Basis of Mental Content.”

Loar, B. (forthcoming b). “Transparent Experience and the Availability of Qualia.”

McGinn, C. (1989). Mental Content, Blackwell.

McGinn, C. (1991). The Problem of Consciousness: Essays Towards a Resolution, Blackwell.

Pitt, D. (forthcoming). “The Phenomenology of Cognition, or: What is It Like to Think that P?”

Putnam, H. (1981). Reason, Truth and History, Cambridge University Press.

Searle, J. (1992). The Rediscovery of the Mind, MIT Press.

Siewert, C. (1998). The Significance of Consciousness, Princeton University Press.

Stich, S. and Warfield, T., eds. (1994). Mental Representation: A Reader, Blackwell.

Strawson, G. (1994). Mental Reality, MIT Press.