As of April 2nd (HK)

Holistic Knowledge

Hilan Bensusan
University of Brasilia
Manuel de Pinedo
University of Granada

Donald Davidson inaugurated an extremely promising and fruitful way of conceiving the way thought responds to the world. The idea is to attempt to show that intelligibility and responsiveness to the world are closely tied. A critical mass of thoughts that is large enough to be semantically self-standing – does not require the contribution of any further thought to be understood – has to be in contact with the world. Davidson’s original insight contrasts with the more commonly held idea that there should be specific points where thought makes contact with the world. Often these specific points are singular items that are deemed immediately accessible (for example, by experience). The availability of specific points of contact is a central tenet of what we shall label ‘the bottleneck picture of how thought connects to the world’. In this paper we intend primarily both to debunk this picture and to draw some consequences of a thoroughly holistic picture of human knowledge.
We draw on Davidson’s holistic conception of contact with the world to develop an account of epistemology that enables greater room for research in ontology. The bottleneck picture has distorted our ontological imagination by forcing us into taking part of the world as more readily accessible. Further, a non-holistic epistemology encouraged ontology to make room for a distinction between matters of facts and what can be established without consulting the world. Such a distinction contrasts directly with the holistic picture we favor. Less constrained by these epistemological requirements, we sketch a combination of holism with an ontology of powers that understands causation, properties, dispositions and singularities in a way that contrasts with the bottleneck picture.

1. The Bottleneck Picture

The idea that we need a bridge between thought and world often springs from the assumption that thinking could be intelligible while utterly indifferent to the world. It follows from the assumption that thought cannot respond to the world unless there is a specific site within it dedicated to receive external constraints on what we think. Such a site is then understood as the part of our thinking that receives messages from the world: the bottleneck through which the world is inserted in thinking. Bottlenecks are often related to exercises of receptivity––as opposed to the mind’s spontaneity, in Kantian terminology (see KrV B75). As such, they are understood as a passive disposition in our mental life: where thought receives (specific) imprints coming from the world. Something outside thought leaves its mark on thought by affecting the bottleneck. Further thought can elaborate a worldview but the credentials for such spontaneous exercises would ultimately lie on what is directly received through the bottleneck.




Perceptual experiences are often seen as suitable to provide bottlenecks. Not anything in the world can be directly in contact with our thoughts––only that, for example, that is available to our perceptive capacities––and not any thought can be directly responding to the world––only, for example, the ones directly connected to our senses. In the general case, the picture holds that contact with the world ought to happen through a bottleneck; through a channel that is narrower than thought itself. Bottlenecks are required because without the specific messages provided by a dedicated contact channel (be it one made of sense detectors, or basic foundational beliefs or passive experiences) our thought would lack the link that makes it possible for it to be affected at all by the world. Botlenecks are bridges connecting (otherwise isolated) thought to the rest of the world. Bottlenecks are typically––yet not always, as we shall see––composed of intermediary objects between thought and world. Empiricism, for example, maintains that the bottleneck is made of sensations (or sense-data, or nonconceptual content, or pure qualia) which would be readily available to us independently of any conceptual exercise or inferential capacity––they were given.
The appeal to a bottleneck is nevertheless independent from any commitment to the dualism of conceptual scheme and empirical content diagnosed and criticized by Davidson (1974). Perceptual content could be intertwined with conceptual capacities and thought could be informed about the world through a conceptually laden and yet passive sphere of our mental life, as in the understanding of experience that McDowell recommends; experience still can act as a bottleneck even though no intermediaries are postulated since any conceptual content could be a perceptual content. We can find the opposite position in Quine: empirical content can be separated from conceptual schemes but there is no bottleneck through which specific messages about the way the world is are carried––sensorial stimuli inform the whole of our thought. At this point we can formulate the bottleneck image of our responsivenes to the world:

Bottleneck Picture: our thought receives specific messages from the world through a dedicated passive channel.

The bottleneck constitutes the region of our thought where there is no activity but mere passive reception of specific messages. The picture has therefore two conjoined components: the appeal to a dedicated element of receptivity in our thought and an atomism about our responsiveness to the world.
Now, the bottleneck picture would seem compulsory if we find no alternative way to satisfy our cravings for rational contact with the world. We would then feel forced either to postulate a bottleneck through which messages from the world are received or to accept that thought is completely disconnected from how things are. Problems with the picture arise from the guiding assumption that thought can be intelligible while being devoid of any constraint from how things are if a bridge with the world is not somehow added. We are left with the option of a bottleneck or no contact whatsoever; if thought is fully intelligible with no appeal to how things are, either thought can be utterly indifferent to the rest of the world or there should be a specific connection between thought and world that needs to be included in the picture. This reasoning is what makes place for bottlenecks.
We find two serious drawbacks with the bottleneck picture. First, once the bottleneck is put into question—as the skeptic would find a way to do—thought would be disconnected from the world. The postulation of a bottleneck invites the idea that large parts of our thinking are self-standing without any constraint from the world: our world-views would be fully uninformed by how the world is if only the bottleneck is severed. Both the bottleneck picture and the skeptical attacks depend on the guiding idea that our mental life could be oblivious to how the world is: minds function as independent variables. This assumption makes it reasonable to think that (at least a sizeable amount of) our beliefs are formed and justified independently of any truth about the world. If we place the bottleneck in our perceptual judgments, we can conceive of our nonperceptual beliefs (the off-bottleneck ones) as confinable in themselves if only we replace the perception channel by a suitably crafted ersatz (for instance, a hallucination channel). The bottleneck picture encourages us to distinguish between the content of our thoughts and the influence of the world; thoughts and empirical (or world) contents are taken to be detachable: it is intelligible to have the former without the latter. Our diagnosis is that the picture springs from the same source that fuels the idea that a separable part of our knowledge could be labeled a priori and is intelligible with no reference to the world and independently from what is true. Accordingly, if we reject the idea that (part of) our conceptual exercises could be played without any constraint from the world, the craving for a bottleneck could begin to subside.
The second problem is that the picture invites the acceptance of the two dogmas denounced by Quine. If thought is in contact with the world through a bottleneck, the world provides specific verdicts regarding the truth of our beliefs through the bottleneck. This is possible if we take these received contents as fixed, atomistic and intelligible without any appeal to other beliefs we hold. This fixed structure of beliefs and meanings could be provided if we understand postulated connections between concepts as immune to the influence of the world––something like analytical judgments are often taken to be. By contrast, we take the crucial lesson from Quine’s rejection of the dogmas to be that any attempt at establishing a contrast between statements about the world and the world itself makes sense only against the background of other statements about the world. If this is so, the world cannot provide specific messages to our thinking without appeal to the resources available to the rest of our thinking: messages are intelligible only within a framework and, unless meanings can be provided from a separate sphere of analytical judgments, messages ought to be understood only in connection to other beliefs held. If the above lesson is drawn from Quine’s rejection of the dogmas, there is no space left for bottlenecks providing specific contents about how things are. The bottleneck picture depends more on a dualism of belief and meaning than on the dualism of scheme and content: it allows for the world to be conceived as a fixed structure that imposes itself as empirical content through specific messages in a bottleneck; this would require only that we have means to understand these messages no matter what else we believe about the world. In other words, if we take meaning to be independent from our framework of beliefs, we can undestand conceptual content coming from the world through the bottleneck as specific messages to inform our thinking about how things are. If there is no sphere of independent meanings––unrelated to beliefs––these messages would be relative to the beliefs held by the thinker. It therefore follows that a suitable rejection of a principled dualism of belief and meaning would make the bottleneck metaphor unavailable.[i] Of course we can hold some of our thoughts fixed in order for the world to produce a verdict about those thoughts that we place in the bottleneck. We can then either point at this bottleneck, built by fiat, and take it to quench our craving for a connection between thought and world or insist that any bottleneck would be equally acceptable. The former path would make the bottleneck arbitrary and therefore thwart our quest for a place where thought responds to the world: given the arbitrariness of the bottleneck, we might as well just decide that the world is reachable by thought. The latter path, by contrast, would just renounce the strategy of placing in the bottleneck the region where contact takes place: we ought to look for it someplace else.

2. Responsiveness to the world without bottlenecks

The bottleneck image is motivated by the idea that thought in itself can be intelligible while failing to respond to the world: any amount of thought contents could be completely devoid of empirical content. Thought can therefore run thoroughly uninformed concerning how things are. Intelligibility and responsiveness to the world appear as two independent variables that are (contingently) present when we have a piece of knowledge about the world. There is always the possibility that any amount of thought is immune to the world; this possibility opens the space where the skeptic moves. Because matters of intelligibility can be utterly alien to how things are we need a bottleneck to provide constraint from the world, otherwise our thought would run in complete indifference towards how things are.[ii] If we find a way to challenge the independence of intelligibility and responsiveness to the world, we can discard both the need for a bottleneck and the impression that the world would be lost without it.
We believe Davidson has at least sketched an account of intelligibility according to which it is a necessary condition for a sufficiently large class of thoughts to be intelligible that it responds to the world.[iii] We can begin to come to grip with this account by considering some skeptical challenges and some usual ways to overcome them. A frequent aim of skeptical argument is to sever the connection between our thoughts and the world: what seems to be a good path to attain truths, the skeptic suspects, lacks, in fact, proper credentials. Credentials are missing because each of our beliefs relies on other beliefs and the possibility is open that our beliefs could all (or most) be false. The skeptical challenge can be summarized in terms of an argument of illusion[iv] (AI) that can be presented as follows:

(P) (It is intelligible that) I can be wrong about each of my beliefs.

(C) Therefore, (it is intelligible that) I can be wrong about all of my beliefs.

One can accept the inference and attempt to show that premise P is false. That has been the effort of most epistemological endeavours and often this is done by arguing that not anything can be wrong––seeking foundations that cannot be wrong. Most (if not all) foundationalist projects rely on the bottleneck picture; even though bottlenecks are not always foundations. Once accepted the inference, the conditions are present for taking intelligibility and responsiveness to the world to be independent and all is set for the bottleneck picture to look compulsory. If we can understand our beliefs while considering that all of them can be false, we will feel compelled to either accept a skeptical conclusion or to look for a way to ensure us that, for some reason, not all of our beliefs can be false and therefore P is false.
However, we can also resist conclusion C by showing that the inference is not valid while accepting P. This would take one to a different way to understand intelligibility. If each of my beliefs can be wrong but not all of them, then not every belief can stand (or fall) on its own. (If one rejects the dogmas denounced by Quine, this is indeed what one expects.) At least some beliefs require other beliefs both to saddle them with content and to promote a contrast between them and the world. The line of argument to the effect that AI is invalid has to take beliefs as items that cannot be understood or individuated away from the critical mass of thoughts where they belong. We can try to present this line of argument (call it Anti-AI) in terms of a critical mass of semantically self-sufficient set of thought contents (an S4 for short). An S4 is a set of thought contents that can stand (or fall) on its own––it follows from what we have already said about this line of argument that at least most of our beliefs cannot be on its own in an S4. Surely, an S4 is not just a set but it should be organized in a network-like format; we shall nevertheless refer to this organized network of thought contents––the argument can also be presented in terms of beliefs––as a set since we shall not explore much the network-like features of the elements in the set. The Anti-AI line of argument goes as follows:

CMT-I: A thought content cannot be intelligible on its own. It requires other thought contents to saddle it with content (Critical Mass Thesis concerning Intelligibility)

CMT-V: A thought content cannot face alone a verdict from the world as it cannot be understood on its own. It requires other thought contents to make it possible to confront it with the world. (Critical Mass Thesis concerning Verdicts from the world, follows from CMT-I)

HT: There is no principled way to isolate the thought contents that are responsible for our understanding of an S4––there are no thought contents that can be taken to play a purely semantical role in the set. (Holist Thesis)

UMET: There could be no resources to confront all (or most of) the thought contents in the S4 one by one and consider that each of them lacks contact with the world. Confrontation between thought and world can only happen within an S4. The diagnosis that all (or most) thought contents in the S4 lack contact with the world is unintelligible––and therefore cannot be made. (Unintelligibility of Massive Error Thesis, follows from CMT-V and HT)

CONCLUSION: Each thought content in an S4 can fail to be in contact with the world––but all of them cannot. A (sufficiently large) part of the S4 has to be in contact with the world but we cannot specify which thought contents are those without isolating them from the others and that would amount to make their confrontation with the world unintelligible.
If the theses are acceptable and the inference is valid, this can provide an argument againt AI. CMT-V and UMET are motivated by the idea that confrontation of thought and world cannot take place when there are no sufficient beliefs making the contrast possible from the beginning. Now, if Anti-AI works, thought has to be understood as coming in collections and such collections––if intelligible at all––are in contact with the world even though we could not specify which thoughts are making this contact possible. For any S4, it will respond to the world if and only if it is intelligible. If Anti-AI can be substantiated and it can be shown that the whole set of thought contents of a given person at a particular moment is an S4, then we will have proven that thought cannot fully fail to respond to the world. Moreover, it shows that neither experiences nor the knowledge about the contents of one’s own mind can sustain independently the contact between thought and the world.[v] Contact with the world can be ascribed to the whole S4 but not to any item in it. As a consequence, the world appears not as external to thoughts but rather as part of what makes them intelligible.
If Anti-AI is sound and manages to engage both genuine thought and world in a way that imbrincates them, a fairly general argument against (at least) some forms of global skepticism can be presented: thought cannot be both intelligible and massively false. The scheme of argument is therefore addressed to (some forms of) global skepticism that would share with the bottleneck picture its motivating assumption concrening thought: responsiveness to the world and intelligibility are independent variables.
In order for the line of argument to be applied we need to specify first what counts as an S4. A first candidate, as mentioned above, is the set of all thoughts a person entertain in a certain moment of time.[vi] This S4 is the one that is used in the argument against global skepticism: one’s beliefs cannot be both massively false and intelligible at the same time. One could ask further whether more restricted sets such as the set of all of one’s beliefs about the future (or all my beliefs about myself, or all my thoughts about the external world) are suitable candidates. If they are, stardard forms of local skepticism (concerning the future, the external world) and challenges to my authority over my own mental life could be also straighforwardly rebutted by instances of the scheme.[vii]. These sets, nevertheless, can hardly be self-sufficient as it is arguable that beliefs about the external world depend for their meaning on beliefs about our language and our practices, beliefs about myself depend for their content on the world or beliefs about the future can only be understood under the light of other beliefs about time. We could make them semantically self-sufficient by adding some meaning postulates to each set of beliefs, but this would violate HT.
Anti-AI, however, can indirectly provide resources to deal with cases of skepticism of more restricted scope. We can see this if we consider two properties of an S4 that we can call properties of self-dependency. First, an S4 cannot be understood without the concourse of the beliefs inside it; it cannot be understood without (at least some of) its items being understood. Second, an S4 cannot be specified without the concourse of beliefs inside it; we can only specify it with the aid of its elements. Now, take the set of my beliefs about the future (F). These beliefs cannot be all false because if so the idea of future events would be meaningless and therefore skepticism about all inductive inferences concerning the future would be wrongheaded (see McDowell 1995). In order to determine what are the beliefs that belong to the set of beliefs about the future we need to make use of some of those beliefs––in order to doubt whether the sun will rise tomorrow we need to hold true that what we call “tomorrow” is determined by, say, a clock ringing at midnight. F cannot be fully understood from outside it––that is, making use only of the resources provided by beliefs that are not in the set. Even though F is not an S4, it enjoys the two properties of self-dependency above and cannot be understood (nor can it be specified) without the concourse of the beliefs within it. We can say it is a semantically self-dependent set of thought contents (SSDS for short).
Clearly, every S4 is also a SSDS­­. Also, once we remove all my beliefs about the future from the set of all my beliefs (T), we obtain a set T – F which could not be understood (or specified) without the concourse of the beliefs within it. T – F is an SSDS but not an S4: we need beliefs about the future to understand parts of our vocabulary related to the past and the present. Skepticism about all beliefs about the future (F) is not granted because F is close enough to an S4 (without being an S4) not to be understandable (and specifiable) without appeal to its own items; and this is so because it is an SSDS. We believe it is reasonable to extend this conclusion to all forms of skepticism that range over SSDSs C such that T - C is not an S4. In contrast, beliefs about my cousin’s car could be removed from T without the remaining set ceasing to be an S4. That makes this skepticism less challenging as T can be regenerated from the removal of all my beliefs about my cousin’s car: it doesn’t cease to be an S4. It can be argued that forms of local skepticism about a set of beliefs C, so that T – C is not an S4, are not granted as we cannot specify and understand either C or T - C without holding true some of C’s items. Therefore, Anti-AI can provide elements to counter any skepticism concerning sets of beliefs that can only be understood and specified using resources within the set.[viii]
Anti-AI can be applied clearly and directly (at least) to the case of all my beliefs at a given time. The line of argument encourages a kind of externalism according to which thought’s contact with the world is intrinsic rather than extrinsic: contact has to do with intelligibility. In “Externalisms” (2001), Davidson considers two forms of externalism––perceptual externalism and social externalism––and recomends a combination of both (through triangulation). In both cases the origin of thought is traced back to some external conditions: we need other people to inculcate in us concepts and criteria or we need causal connections with the world in order to have something perceived. These elements of the world (elements ouside what is available to the thinker) are, in an important sense, external to a first-person awareness of her own thinking: an argument can be made that awareness of one’s own thoughts requires no awareness of the elements of the world that enable those thoughts to take place. One’s thinking, as far as first-person awareness is concerned, could be completely oblivious to any (true) theory about what makes thought possible. These externalisms, in a sense, only add more to our third-person theory of the world––they make the link between thought and world external and therefore just add a further (postulated) element to the (assumed and perhaps true) furniture of the world.[ix]
Anti-AI, however, insinuates a way of making thought stand in contact with the world in an intrinsic manner; that is, in a way that is not oblivious to first-person awareness. It suggests an externalism where the connection between contact with the world and intelligibility makes access to the world a feature of thought; it is an intrinsic feature of thinking that the world is within its reach. It therefore adds no further theory about the origin or the nature of thoughts––no theory that expresses a third person, metaphysical or external dependence between thought and world. The externalist picture favoured is rather that thought depends on some access to the world: the world enters the picture as the object of one’s thought and not as something postulated from outside what is within the reach of the thinker. Accordingly, as thought responds to the world because of one of its intrinsic features there is no need to appeal to bottlenecks connecting thought contents with how things are.
Davidson has been sometimes charged with an accusation of simply postulating that there are thoughts. Peter Klein (1986), for one, has claimed that Davidson never argued that there are such things as coherent beliefs about the world––their existence was never more than merely taken for granted. Klein’s emphasis on beliefs about the world might seem misplaced as the world is gained in the argument through a firm connection between truth and intelligibility. The gist of the charge, however, is that if there are no thoughts, an argument like anti-AI cannot respond to the skeptical challenge that we have no knowledge. The suspicion is that Anti-AI does no more than postulate thought contents and explore their features. Surely, if there are no thought contents, there could be no responses to the world: there is nothing for the world to act upon. In that case, the skeptical case cannot be presented. The case against AI, then, can be no stronger than this conditional formulation: if there are thoughts, anti-AI holds. This, however, could be enough for most purposes––in particular to ensure a link between thinking and responding to the world. Davidson has often recommended the view that thought contents are products of interpretation––thought cannot be detected but within the borders of thought. Inversely, having thought consists on detecting thought on oneself and others. What Davidson takes for granted is that there are instances of such detection, which could not be understood if there were no thought contents. It is unclear how things would be if there were no thoughts: our constant endeavour to interpret each other could be enough to make sure that we can ascribe thought––and if so, we can ascribe knowledge.
The strategy to understand responsiveness to the world in terms of some features of thought could give the impression that we are settling for something somehow weaker than full-blown contact with the world. This worry attends to the echos of Kant’s transcendental distinction between things-for-us and things-in-themselves. The distinction has close resemblance with a persistent global skeptical challenge: anti-AI has not ruled out the possibility that the world is entirely different from whatever we can think about it. Of couse, we could not confront the world with our thoughts (because of CMT-V) but it cannot entirely escape our scrutiny. Surely nothing could be said as to how the world is. Anti-AI, suggesting that our thoughts could not be entirely off the mark, ensures us that even though any bit of the world can fail to be correctly thought, some bit of the world has to be grasped by thought. Anti-AI provides then with a two part argument against challenges like Klein’s: on the one hand it grants the existence of thought and thought contents, on the other guarantees that those contents are such that at least some of them must accommodate the way the world is.
Anti-AI was presented here in terms of intelligibility and contact with the world. When we claim that a large part of the set (of beliefs) is in contact with the world, we can take beliefs that contact the world, as Davidson often does, as true beliefs. Truth, then, would be an element that has to be present in an S4; even though we cannot isolate which elements in the set are true. We can take most of them to be true or rather claim that there is a presumption of truth in favour of any belief within an S4 (Davidson 1983). Truth of some beliefs is what is required to make an S4 understandable. Truth, as Davidson has often pointed out, has an intimate connection to understanding and interpretation. Intelligible beliefs are those that can be understood and considered against a background of true beliefs. Truth, however, is not the only way to understand contact with the world: we can also, for example[x], think that some beliefs in the set have to constitute knowledge. This makes sense especially if knowledge is taken to be the basis of justification (Williamson 2000) and a condition for any judgement concerning truth. A set of beliefs is intelligible only if there is enough knowledge about the world to make a confrontation between each belief and the world intelligible. We cannot, due to HT, isolate where there is knowledge––we can only safely assume, if Anti-AI works, that there is some knowledge disperse within the set. We shall come back to this idea in section 4.

3. Two (non-recommended) approaches to holistic constraints from the world

Our purpose in the remainder of this paper is not to directly assess Anti-AI but rather to examine some of its consequences. We shall distinguish three general images of the relationship between thought and the rest of the world that could be encouraged by Anti-AI and then proceed to elaborate and recommnend the third. The first is Davidson’s way of putting together the connection between truth and intelligibility on the one hand and his ideas about mental content and physical events on the other. The second is a variation of McDowell’s ideas that emphasize the thesis of the partial re-enchantement of nature and takes facts to be somehow constituted by thinkables or thinking contents. The third general image has elements of the first two while avoiding some problems with the second and some perceived drawbacks of the first. In this section we shall sketch the images and point at our discomfort with some of the features of the first two.
The first image is one that provides a holistic account of our mental life and therefore of sets of beliefs while still insisting that each individual mental event be describable in non-mental terms. This picture associates the line of argument in Anti-AI with a number of other theses by Davidson that revolve around anomalous monism and a causal individuation of events. Our beliefs are about a world that interacts with us and thus causes us to have the beliefs we have. Each of our thoughts can be causally in contact with the world and therefore the semantical contact provided by Anti-AI is not the only way by which our beliefs touch the world. The presumption of truth in favour of our beliefs comes together with the thesis that those beliefs are about what cause them. Now the causal connection between beliefs and what prompts them can be described in physical terms as any event can be described in physical terms. The nomological character of causality guarantees that if there is a causal connection between “p” and “belief that p” there is a law connecting them but this law does not have to be presented––and indeed cannot be presented––in terms of beliefs and other mental terms. In fact, Anti-AI, and in particular UMET, can motivate here at least a weak version of the doctrine of the anomaly of the mental: if there were atomistic psychophysical laws––laws that connect individual physical states to individual mental contents––there would be a way to specify which thought content is connected to which physical event and therefore it would be in principle possible to determine at least some thought contents independently of any other thought content. It would then be possible to connect “p” and “belief in p” independently of other beliefs: provided with an atomistic psychophysical law of this kind we would be able to trace an event of the world connected to each belief.[xi] At least this rejection of atomistic psychophysical laws is required to grant sovereignty to mental contents: our beliefs respond to other beliefs and make contact the world in blocks through a connection between truth and intelligibility. Davidson’s anomalous monism also provides an account of events whereby they can always be described in the vocabulary of physics; this allows for causal contact between events that can be described mentally and other events without compromising the nomological character of causality. This causal contact provides a further account of the contact between thought and the rest of the world that is added to the one concerning how truth is required for intelligibility.
Now, it can be useful to distinguish two strands of thought within the body of doctrines espoused by Davidson. On the one hand, in his work on radical interpretation, truth and meaning, he has argued that any exercise of understanding intentional behaviour, linguistic behaviour in particular, must start with a presumption of coincidence between the speaker’s and the interpreter’s systems of beliefs and of general truthfulness regarding those beliefs. Davidson calls this presumption the principle of charity and it appears under numerous guises: his rejection of the scheme-content dualism––and therefore of a principled separation between a thinker’s contribution and the world’s contribution to a system of beliefs––, his defence of triangulation as a precondition for thought––communication happens when speaker and interpreter compare their perspectives on a common world––, the interdependence of the three varieties of knowledge––about oneself, about others and about the world––and his truth-conditional account of meaning––meaning, belief and world are interdependent. Also his (often controversial) thorough rejection of a public language (1986) can be made to somehow fit into this strand of connecting intelligibility and access to the world through interpretation. This first strand can be seen as motivating and giving flesh to Anti-AI.
A second strand of thought, from which we take the first to be independent (and arguably incompatible), includes his defence of a monism (of an anomalous variety) and his causal account of the contact between beliefs and the world, together with his work on the individuation of events. Adding this second strand to the picture, mental events––normative and holistic as they are according to the first strand––must also be physical events in order to have causal powers. Our contact with the world through our beliefs can therefore be understood in physical terms––a physical description of how thoughts manage to be rationally constrainted by the world can be provided. Because the nomological character of causality is endorsed, there should be a law connecting physical description of our beliefs and their contents––our thoughts and the world. Such a law makes sure that beliefs are causally connected to the world and, as a consequence, the relation between our mental life and the rest of the world is always of a causal nature. We shall claim that this causal account of the relation between beliefs and their content adds unnecessary difficulties.
Guided by his causal acount of content, Davidson cannot fully preclude the possibiluty that we could determine the content of beliefs without engaging in an activity of interpretation; it would be enough to discover to what our beliefs are (appropriately) causally connected. As a consequence, once it is granted that this alternative way to determine the content of a belief is available, one can be inclined to try and determine the content of a belief without engaging in interpreting it and therefore without understanding it. In fact, according to Davidson, if anything is systematically causing a belief (or an utterance) that is what the belief is about (1990: 201).[xii] If this is so, not only interpretation can be (in principle) rendered dispensible but also there are two potentially conflicting authorities concerning the content of a belief––or, more generally, concerning one’s mental life. The swampman mental experiment proposed by Davidson (1987: 18-19) illustrates the predicament: Davidson concludes that the swampman (an indentically replica of Davidson produced by a cosmic accident simultaneously to the destruction of Davidson by a lightning) has no mental life even though people around him would ascribe to the replica most of Davidson’s beliefs and desires. It is just from the perspective of someone who knows that the accident had happened (and the replica has been brought about) that one could say that there is no mental life going on there because there is no causal link between any state of the replica and the contexts in which Davidson’s beliefs and desires where acquired. This swampman’s absence of a mental life can be appreciated only from a third-person perspective not available to the standard interpreter of the replica’s utterances that meets him after the accident and is not informed of what has happened. If we assign the ultimate authority to this perspective, we make our mental life ultimately dependent on causal relations around us. Surely our beliefs and desire are causally related to the world in different ways but we cannot intelligibly talk about these causal connections outside the perspective of an interpreter. If there is a shortcut that would take us items of one’s mental life without interpretation, we could have beliefs confronting the world without the aid of further beliefs and that would contrast with CMT-V.
Moreover, the idea that there are physical descriptions of beliefs suggests that such events could relate to each other in a one-to-one basis, contrary to HT and to the overall spirit of Anti-AI. Davidson’s account of monism depends also on how we individuate events so that we can establish that a belief and its content are the events mentioned in the corresponding nomological connection. The effort to individuate events took Davidson to present a causal account of sameness of events that run in its own difficulties (see Davidson, 1985). Physical descriptions of beliefs place them further within a causal structure where they can be individuated and treated separately.[xiii] Surely, that alone is not enough to commit Davidson to the bottleneck picture but it adds an unnecessary tension with CMT as it seems to fly in the face of the idea that only critical masses of beliefs can be in contact with the world.
Davidson adds a causal story to the one concerning beliefs being in contact with the world as a block – rational contact with the world through confrontation of critical masses of beliefs. The causal story, however, is not necessary as the first of the two strands above doesn’t require or depend on the second. The first strand, and Anti-AI in particular, can dispense with a causal account of the relation between beliefs and the rest of the world—and dispense with occasional sentences, criteria for individuation of events irrespective of their description, and causal determination of mental content. Surely, as we shall see, there could be causal relations between beliefs and other events but there is no need for a causal channel between individual beliefs and physical events to be postulated. We claim that the rational connection between critical masses of beliefs on the one hand and the world on the other is enough to assure us that we cannot be indifferent to how things are. Anti-AI can but need not be associated to anomalous monism and its commitment to the nomological character of causality and the causal character of the connection of beliefs and their content. The elements of what we have called the second strand in Davidson’s body of doctrines can be dispensed without prejudice to the structure of our account of how critical masses of beliefs respond to the world.
The second image that can fit with Anti-AI can be brought to view if we consider what happens when we retain from the bottleneck picture the idea of responsiveness to facts while letting go of a channel that receives these facts. An alternative is to consider that conceptual capacities are always required in order to detect facts of the world. Rational constraints then can come from the facts absorbed by any of comceptual capacity. This alternative bears similarities to the position recommended by McDowell (1994) but makes no attempt to rehabilitate a (privileged) tribunal of experience; thought is in contact with the world because the world reveals itself through conceptual capacities. In a sense, this alternative endorses McDowell’s partial re-enchantement of nature while rejecting any special role to experience.[xiv] Our thought is not constrained by a nonconceptual Given while it is not frictionless spinning in the void as conceptual constraints could also be legitimate worldy constraints. Thought is composed by contents that cannot stand or fall on their own and, as conceptually articulated items, can, for example, be identical with items of the world.[xv] Such alternative contrasts with the first one mainly in its conception of the world: the world is all that is the case, a totality of facts while being a fact, something that is the case, already involves being thinkable. Intelligible thinking reaches for facts of the world. Facts and thought contents are understood as having the same structure; that itself is what makes it unmysterious that our thinking could be legitimately respond to the world. The image therefore attempts to present a world of facts fully aprehensible through our conceptual capacities and therefore readily accessible to thinking without a bottleneck connecting our beliefs to how things are.
This image, however, has serious limitations.[xvi] Its central drawback is its failure to accommodate CMT. While we can make sense of the idea of particular causes affecting our thought without the causal relation needing to be subsumible by a general law, there is a tension between the thesis that thought is in nature holistic and the idea that the world is a collection of independently conceivable (conceptually aprehensible) facts. The most famous version of this picture (the one offered in Wittgensteins’ Tractatus) presents an atomistic conception of the world (the world divides into facts and each item can be the case or not the case while everything else remains the same; 1.2-1.21) and thoughts (or the propositions that express them; 3.1) are independent of each other (2.0211) while describing reality completely (4.023). Basic states of affairs (facts that are the case) are taken as independent of each other and composing what Wittgenstein famously considered one of the prototipycal manifestations of the mystical: the feeling of the world as a limited whole, that is, as everything that is the case, as a collection of all the true facts. The picture where there is a structured world of (ready-made) facts contrast with a holistic picture of thought as it imposes on our thinking a structured collection of facts to which we respond––structured facts that are instances of correction or something akin to truth-makers[xvii] for our thoughts. The world is therefore impressing on us facts and our thinking responds to them––the tension between individual facts and a holistic conception of thought shows up again; this time without any appeal to a causal connection between beliefs and their content. It seems that this second alternative has to give away at least some holism as our thinking responds to specific atomistic facts that constitute the world. This commitment with the second dogma denounced by Quine could quickly encourage a separation between the contribution of the world to our thinking––that can be mapped into the facts that the world itself provides––and our own additions where our thinking would be up to us only. A distinction between meanings and beliefs seems to be therefore strongly encouraged––and both HT and UMET seem to be called into question.
A variation of this image can be suggested by a view according to which the Tractatus was not committed to a world of ready-made facts.[xviii] If this is so, the talk of atomistic facts could mean either that these facts are partly constituted by something structural to our thinking or that they respond to our process of thinking––and therefore are revisable on the light of the rest of our thinking. The former alternative would still be prey to the difficulties rehearsed above. If there are fixed ways in our thinking about the world (even if they don’t come from the world itself but from our interaction with it), then part of the holistic character of thinking has to be compromissed and we can feel invited to embrace a version of the distinction between belief and meaning. Whenever facts–– either conceived as part of a fix structure of our thinking or as ready-made in the world––are individuated independently of our process of thinking, there is room for a distinction between facts on the one hand and processes of thinking on the other. The latter alternative above––where facts are individuated not by the world or by some structural element in our thinking but by our thinking itself––would escape our criticisms but would offer no more than a faint suggestion as to how thought manages to respond to the world. Our third alternative will encompass something close to this last suggestion as it avoids individuating facts in any way independent from our sets of thought contents. By presenting this third alternative, we hope to take holism as a feature of mental content that has to be conciliated with our responsiveness to how things are, rather than sacrificed in order to obtain contact with the world.

4. A more satisfactory approach to holistic constraints from the world

A third––and, we shall claim, better––image can come to view as a consequence of the drawbacks we diagnosed in the first two. If contact with the world is a feature of an S4, there is no need to postulate a fixed or prestructured picture of the world; either in terms of causal connections or in terms of individual, atomistic facts. There is no need to have a fixed sketch of an account of how the world we are making contact with should be: our contact with the world is the sole source of knowledge about it. In fact, we shall focus on knowledge and take Anti-AI to entail arguments that our knowledge of the world is to be present in an S4. We could therefore endorse, in a different context, the slogan championed by Williamson (2000): knowledge first. This alternative will take knowledge as a condition for the intelligibility of a critical mass of beliefs: knowledge is what ought to be present in the S4 even though it cannot be separated from the rest of the set of thoughts. It is therefore a holistic conception of knowledge where knowledge is present within a network of beliefs but fails to be a piece of knowledge if taken in isolation­—it cannot, in fact, intelligibly be taken in isolation. Further, while some knowledge has to be present in a critical mass of beliefs, there is no need or means to specify what in the critical mass amounts to knowledge.
Holistic knowledge––as we label this third alternative to stress our focus on the presence of knowledge in an S4––means that we cannot pinpoint what is known while we can infer from the intelligibility of a network of beliefs that there is some knowledge present. As knowledge is our point of departure to reach the world, we don’t need to postulate any definite structure for the world we are contacting. The world emerges in the effort to interpret—render intelligible—a network of thought contents. There is therefore no need to appeal to a fixed structure of known facts or known causal structures––what needs to be known is nothing specific, knowledge is the (potential) starting point for more knowledge.[xix] Knowledge is present and yet disperse within an S4. This third alternative takes knowledge to come before we have an overall picture of the world with which we are to make contact––contact, through knowledge, is present from the beginning and Anti-AI is enough to show that. There is no need to put any metaphysical chart in front of our knowledge horse: we do not have to feel forced to favour ontologies of facts, events, properties or objects, as any of these could be available to a conception of thinking grounded on knowledge—they could, in principle, be known.
We have seen that thoughts, beliefs or experiences are neither intelligible in isolation (CMT) nor can individually ground an S4 (HT). It follows that a knowledge claim cannot be understood as a claim about a specific content, connected to the world independently of other contentful states of the thinker. Knowledge claims are claims about a critical mass of thoughts in a network and can only stand within the critical mass. At the same time, thoughts have contents specifiable within the network as claims and those claims can be taken to be knowable contents[xx], something that can be known; and we shall say no more about the world except that it is constituted by knowable contents. Now, access to knowable contents is achieved, indistinctively, through perceiving, reflecting, feeling or acting, which invites detachment from any clear cut compartimentalization of modes of engagement with the world. No preference is given to any area, perceptual, inferential or practical, of our interaction with the world in coming to access knowable contents.The idea is that access to the world can only take place within our S4: holistic knowledge promotes access to the world through our network of beliefs. The world is made transparent by the capacities for correction of any belief that can be found within a set of beliefs. Even though we cannot detect where there is knowledge, we can detect (holistic) knowledge within an S4 and that knowledge is what entitles us to claim that we are in contact with the world.
The image is designed to add the least possible to what is already in Anti-AI and so to preserve a fully holistic account of mental content. There is an important feature of holism that the image can accommodate. If we take holism seriously, no individual belief is ever shared with anybody for contents are relative to other contents; however, we start interpretation through an application of the principle of charity: we make use of some of our beliefs in order to understand others. This third alternative claims that we do likewise with the world: we understand it from the perspective of our beliefs. Therefore, the world cannot on its own make individual beliefs true but it is part of the knowledge that is present in a network of beliefs; contact with the world requires a critical mass of beliefs. As a consequence, we cannot say that we know p but at most that p belongs to an S4 where there is knowledge. And, of course, we can argue for p with the resources available in the S4. In a sense, the world tells us something only in the context of interpretation––knowledge springs from a basis of intelligible thought. Beliefs, and knowledge, can only be understood collectively; our contact with the world has to be therefore present (and maybe massive) for us to start thinking. Thinking depends on a critical mass of beliefs but that, in itself, is what grants us contact with the world­­––that is, according to this alternative, what grants us knowledge. Knowledge is a claim about a critical mass of thoughts. What a knowledge claim is a claim about can only be established against the background of knowledge that entitles it to its content––we can specify, at best, knowable contents. Beliefs can surely be contrasted individually but that is done only against a background of intelligibility. If this third alternative is acceptable, as we believe it is, the very idea of intelligibility––and its connection to responsiveness to the world–– is all we need to ensure that our thinking involves holistic knowledge of the world.
One can wonder how a holistic conception of knowledge could cope with issues related to justification and the goal of inquiry. It can be claimed that, if any S4 has (holistic) knowledge and we cannot determine when an individual belief that we have adquired is in fact knowledge, inquiry needs to be explained in terms other than seeking knowledge. Now, within a network of beliefs there are justification relations and those can be assessed. We could then have networks of beliefs with different degrees of justification and can formulate the goal of inquiry in terms of justified beliefs (maximize justified beliefs, minimize unjustified beliefs, etc). Of course, we cannot define or understand knowledge in terms of justified beliefs; we have at least to make room for increases in justified beliefs that would not amount to more knowledge. Justification takes place in an environment that is already laden with beliefs and knowledge-rich.
The presence of knowledge is what instills contact with the world in an intelligible network of belief––justification is a way to exploit that contact with the goal of having a network of beliefs with greater rational responsiveness to the world. The pursuit of justified beliefs in this context fulfills the purpose of securing safer subnetworks of beliefs (theories, worldviews). There is no reason to think this is not an achievable and worth-pursuing purpose. But safer subnetworks of beliefs––where we can be sure to find a greater amount of knowledge––are only relative to our broader network of beliefs that itself has to host a measure of knowledge. Holistic knowledge, in contrast with justified beliefs, is elusive. Further (holistic) knowledge can only be diagnosed by arguments that consider the network of beliefs as a whole and draw conclusions from how justified the beliefs within it are. Justification is always justification with respect to a network of beliefs and therefore with respect to some holistic knowledge. Intelligibility is itself connected to justification: interpretation starts always from a network of beliefs where holistic knowledge is present. Beliefs themselves are connected to knowledge through inteligibility and interpretation. We seek justification because we want to ensure ourselves a less elusive contact with the world. Our attempts to proceed an inquiry starts out from a measure of knowledge while no specific piece of knowledge can be presented––there is no principled way to tell apart pieces of knowledge from other beliefs of the network. The best we can do is to seek justified beliefs that seem firmer than others while knowledge itself is not guaranteed but within the network as a whole.
Holistic knowledge is incorrigibly elusive. The emerging idea is that we can provide no resources to produce a set with nothing but knowledge. We prefer to talk about (holistic) knowledge to emphasize the elusive character of our contact with the world: we cannot pinpoint where we touch it without presenting a bottleneck. Holism can be conciliated with contact with the world only if we give up the idea that we should be able to indicate where this contact takes place. Holistic knowledge is elusive because it cannot be presented in isolation from a network of beliefs. We have means to ensure ourselves that we know about the world––we just have no means to determine what exactly we know about it.

References

Brandom, R. (1994), Making it Explicit, Harvard, Mass., Harvard University Press.
Brandom, R. (1995), ‘Knowledge and the Social Articulation of the Space of Reasons’, Philosophy and Phenomenological Research, 1995, 895-908
Brandom, R. (1998a), ‘Perception and Rational Constraint’, Philosophy and Phenomenological Research, 58, 369-374.
Brandom, R. (1998b), ‘Insights and Blindspots of Reliabilism’, in Articulating Reasons, Cambridge, MA., Harvard University Press, 2000, pp. 97-122
Brandom, R. (1999), ‘Varieties of Pragmatism: Synthesizing Naturalism and Historicism’, in: R. Brandom (ed.) Rorty and his Critics, Oxford, Blackwell, pp. 156-182.
Brandom, R. (2002), ‘Non-inferential Knowledge, Perceptual Experience, and Secondary Qualities: Placing McDowell’s Empiricism’, in Nicholas H. Smith (ed.) Reading McDowell, London, Routledge, pp. 92-105.
Brandom, R. (2006), Between Saying and Doing: Towards and Analytic Pragmatism, John Locke Lectures, Oxford, May-June, 2006.
Burge, T. (1988), ‘Individualism and Self-Knowledge’, Journal of Philosophy 85, 649-63.
Davidson, D. (1974), ‘On the Very Idea of a Conceptual Scheme’, in Inquiries into Truth and Interpretation, Oxford, Clarendon Press, 1984, pp. 183-98.
Davidson, D. (1982), ‘Empirical Content’, in Subjective, Intersubjective, Objective, Oxford, Oxford University Press, 2001, pp. 159-176.
Davidson, D. (1983), ‘A Coherence Theory of Truth and Knowledge’, in Subjective, Intersubjective, Objective, Oxford, Oxford University Press, 2001, pp. 137-153
Davidson, D. (1985), ‘Reply to Quine on Events’, in Essays on Actions and Events, Oxford, Oxford University Press, 2001, 305-312.
Davidson, D. (1987), ‘Knowing one’s own mind’, in Subjective, Intersubjective, Objective, Oxford, Oxford University Press, 2001, pp. 15-38.
Davidson, D. (1990), ‘Epistemology Externalized’, in Subjective, Intersubjective, Objective, Oxford, Oxford University Press, 2001, pp. 193-204.
Davidson, D. (1991), ‘Three Varieties of Knowledge’, in Subjective, Intersubjective, Objective, Oxford, Oxford University Press, 2001, pp. 205-220
Davidson, D. (2000), ‘Truth Rehabilitated’, in R. Brandom (ed.) Rorty and his Critics, Oxford, Blackwell, pp. 65-74
Davidson, D. (2001) ‘Externalisms’, in Kotakto, P., Pagin, P & Segal, G. (eds.) Interpreting Davidson. Stanford, CSLI, pp. 1-16.
Hornsby, J. (1997), ‘Truth: The Identity Theory’, Proceedings of the Aristotelian Society 97, pp. 1-24.
Kant, I. (KrV), Critique of Pure Reason, London, Macmillan, 1929; trans. of Kritik der reinen Vernunft (Riga, Hartknoch) by N. Kemp Smith.
Klein, P. (1986), ‘Radical Interpretation and Global Skepticism’, in LePore, E (ed.) Truth and Interpretation – Perspectives on the Philosophy of Donald Davidson. Oxford, Blackwell, pp. 369-386.
Malpas, J. (2005), ‘On Not Giving Up the World: Davidson and the Grounds of Belief’,in Smith, P. J. (ed.), Significado, Verdade, Interpretação: Davidson e a Filosofia, São Paulo, Loyola, pp. 1-17.
McDowell, J. (1977), ‘On the Sense and Reference of a Proper Name’, in Meaning, Knowledge, and Reality, Cambridge, Mass., Harvard University Press, 1998, pp.171-98.
McDowell, J. (1979), ‘Virtue and Reason’, in Mind, Value, and Reality, Cambridge, Mass., Harvard University Press, 1998, pp. 50-73.
McDowell, J. (1982), ‘Criteria, Defeasibility, and Knowledge’, in Meaning, Knowledge and Reality, Cambridge, Mass., Harvard University Press, 1998, pp. 369-94.
McDowell, J. (1984), ‘Wittgenstein on Following Rule’, in Mind, Value and Reality, Cambridge, Mass., Harvard University Press, 1998, pp. 221-62.
McDowell, J. (1994), Mind and World, Cambridge, Mass., Harvard University Press.
McDowell, J. (1995), ‘Knowledge and the internal’, Philosophy and Phenomenological Research, 55, 877-893.
McDowell, J. (1998), ‘Having the World in View: Sellars, Kant, and Intentionality’, a revised version of the Woodbridge Lectures; Columbia University, April 15, 16 and 17, 1997, Journal of Philosophy 95, 431-91.
McDowell, J. (2000), ‘Scheme-content Dualism and Empiricism’, in The Philosophy of Donald Davidson, ed. by L.E. Hahn, Illinois, Open Court, pp. 87-104.
McDowell, J. (2002), ‘Responses’, in Nicholas H. Smith (ed.) Reading McDowell, London, Routledge, pp. 269-305.
McKinsey, M. 1991, ‘Anti-individualism and privileged access’, Analysis, 51, pp. 9-16.
Moran, R. (2001), Authority and Estrangement, Princeton, Princeton University Press.
Peirce, C. S. (1868) ‘Questions Concerning Certain Faculties Claimed for Man’, Journal of Speculative Philosophy, 2, pp. 103-114.
Pinedo, M. (2006), ‘Anomalous Monism: Oscillating between Dogmas’, Synthese 148, 19-97.
Putnam, H. 1975, ‘The Meaning of ‘Meaning’ ‘, in Mind, Language and Reality. Philosophical Paper, vol.2, Cambridge, Cambridge University Press, pp.215-71.
Rorty, R. (1986), ‘Pragmatism, Davidson, and Truth’, in Lepore, E. (ed.) Truth and Interpretation, Blackwell, pp 333-355.
Rorty, R. (1999), ‘Response to Ramberg’, in R. Brandom (ed.) Rorty and his Critics, Oxford, Blackwell, pp. 370-377.
Williams, M. (1996), Unnatural Doubts, Princeton, Princeton University Press.
Williams, M. (1996a), ‘Exorcism and enchantment’, The Philosophical Quarterly, 46, 182, pp. 99-109.
Williamson, T. (2000), Knowledge and its Limits, Oxford, Oxford University Press.
Wittgenstein, L. (1953), Philosophical Investigations, Oxford, Blackwell; trans. of Philosophische Untersuchungen by G.E.M. Anscombe, edited by G.E.M. Anscombe and R.Rhees.
Wright, C. (2002), ‘Human Nature?’, in N. Smith (ed.) Reading McDowell, Routledge, pp. 140-73.


[i] We take Quine to hold that no specific message is received from the world; there is empirical content but that content provides no conceptual message as how things are on its own: it requires the intervention of our conceptual sovereignty.
[ii] The idea that thought could be intelligible while indifferent to the world has been challenged recently by the debates on (content) externalism inaugurated by Putnam (1975). If, however, externalism means only that the world needs to be consulted in order to individuate and identify thought contents, the thinker could still fail to believe that her thought can be in contact with the world: there is still room for a skeptical thinker. In order to curtail this possibility, externalism has to make sure that whatever contact with the world that is required for thought is somehow available to the thinker. More on content externalism and what is available to the thinker below.
[iii] The account can be found in many of his works but in more explicit form in 1982, 1983, 1991.
[iv] This line of thought that makes use of the argument of illusion to present some general features of the skeptical challenge is suggested by Davidson himself (for example 1988: 45) and, to some extent and in a different context, by McDowell 1995.
[v] The idea that contact with the world is a task for part and not the whole of our set of beliefs is present in different traditions of thought. Empiricists claim that the realm of experience would take the task whereas some rationalists would rather attribute it to the contents of one’s own mind that are seen as capable to sustain independently the contact between thought and world.
[vi] There is a number of ambiguities in this phrase, mostly to do with what counts as an individual thought and whether a person’s thought is closed for logical consequences (i.e., if the logical consequences of an item in an S4 has to be in the S4). We believe the line of argument can work for our purposes irrespectively of how one goes concerning those issues.
[vii] One can claim, for instance, that great part of my beliefs about my own mind have to be in contact with the world––that is, with how my mind really is––so that my authority over my own beliefs is (at least partially) guaranteed. An S4 of first-person beliefs would imply that “I believe there is a dagger in front of me” has its meaning no matter what “daggers can cut your skin” means; it probably depends only on the meaning of “I believe that daggers can cut your skin”. Now, this recoil to some sort of narrow (first-person) meaning or content is the classical first step towards asserting first person authority––it makes my authority over my own mental states oblivious to how the world is. First-person narrow contents are typical posits of a conception of thought and content where these are private. Richard Moran (2001), among others, presents arguments showing that first-person authority does not depend on first-person beliefs being in any sense private.
[viii] The idea behind an SDSS-based argument against restricted forms of skepticism can be presented in terms akin to those used by Brandom (2006) to suggest an analytic pragmatism that endeavours to elucidate relations between vocabularies and practices. Brandom introduces an apparatus of {sufficient} and necessary relations between different vocabularies and practices––where vocabularies, for example, specify, pressupose or express practices while those are enabling conditions for the usage of (other) vocabularies. Skeptical challenges can perhaps be understood in {terms} of a practice of challenging portions of a vocabulary. Our argument can be seen as showing that the challenge cannot work if the skeptical practice requires, in order to be specified, resources that can only be found in the portions of vocabulary that the challenge aims to put in question. Whenever this is the case, the portion of vocabulary being challenged constitutes an SDSS.
[ix] McKinsey (1991) argued that if the link between thought and world is merely a metaphysical link there would be no risk of one’s having to know about the world in order to know about oneself. The idea is that no a priori knowledge of contingents matters of fact about the world could be preserved. See also Burge (1988). As it will be clear later, we believe that the link cannot be a merely metaphysical link and, as a consequence, part of what we usually claim about a priori knowledge and contingent matters of fact about the world will have to be reexamined.
[x] Instead of truth one can also think of contact in terms of inference validity within the S4. Validity, here, would have to be construed more broadly than usually so that we can talk of valid material inferences. The idea, then, would be that most material inferences enabled by those beliefs are valid. We can claim that there is a presumption of validity in favour of any inference because without the background of shared inferences that we use to understand each other no interpretation could be possible. Inference validity, on the other hand, is what assures contact with the world as we cannot possibly make sense of a confrontation between our thinking and the world that would not take for granted the validity of (great part of) the inferences of which we make use. This inferentialist semantics differs from that recommended by Brandom (1994) as Brandom’s inferentialism requires reliable observational judgments to provide the contact between our inference moves on the one hand and the world on the other . Our alternative inferentialism has that most thoughts are products of valid inference. It seems that such inferentialism might have been suggested by Peirce in his 1868.
[xi] With holist psychophysical laws we would have a connection between a physical event and a global mental life. Even though this would still enable one to bypass the effort to interpret a set of beliefs in order to understand it, psychophysical laws like those don’t seem to be a challenge to UMET.
[xii] He then has to find a way to argue against the charge (aired by Rorty (1986) and McDowell (1994: 16-17) that a brain-in-a-vat would have wires as the content of its beliefs. Even if Davidson has resources to counter the charge (see Malpas 2005), he might be unncecessarily inviting the suspicion by adding a causal story (and what we take to be the second strand in his philosophy) to his account of mental content.
[xiii] Effects of this two separate accounts of content can be also found in Davidson’s account of occasional sentences (see 1983: 152). Davidson’s attempt to conciliate the two accounts can also be seen as in contrast with his rejection of the scheme-content dualism, see Pinedo 2006.
[xiv] McDowell feels encouraged to understand nature as something that is not entirely alien to our conceptual capacities. His route towards this thought starts out from an urge to vindicate the idea of a tribunal of experience after the rejection of the Given as a myth—experience cannot constrain thought without conceptual content. He argues that experience can be both conceptual and passive and further that, being so, it can capture facts of nature. Now, nature is then to be seen as made of facts; conceptually articulated units. We can however dispense with the route and consider only what can be achieved by a conception of nature such as the one McDowell is willing to recommend. McDowell’s position, but not this variation, has to be committed to the bottleneck image. When nature is so re-enchanted, we can passively receive—even if we don’t have experience as a separated place for receptivity—facts about the world. These are ready-made items that can be, at least in principle, taken in one by one. In this context, it is interesting to bring up a diagnosis Crispin Wright makes of McDowell’s position: “[his position] amounts not to a rejection of the Given as such, but a recasting of it. What is given in experience is essentially of the form: that P—that so-and-so is the case. “In experience one finds onelsef saddled with content” (MW, 10). In rejecting the Myth of the Given, McDowell intends to reject a mythology about what is Given, and how, but not the very idea that anything is.” (Wright 2002, p. 145). By re-enchanting nature—no matter whether experience is passive or epistemologically relevant—McDowell proposes a demythologized Given. Now, of course, Sellars didn’t argue, at least according to McDowell, against the very idea that something is received in a (sometimes conceptual) ready-made format by our mind from the world. Sellars’ complaint is against items that get into our mental life without being previously embedded in a theory. Surely then, McDowell’s Given, if it is fair to call it like this, doesn’t fly in the face of Sellars’ efforts to denounce the myth of the Given. However, it goes against the overall autonomy of our body of beliefs to determine the content of each belief—it goes against holism for it assumes that atomic ready-made messages are conceptual contents ready to be taken in by our endeavour to think. The world has to constrain us by telling specific messages (or it cannot constrain us). Quine, for one, can escape this urge to compromise holism by taking the Given to be nonconceptual while taking the tribunal of experience as capable to give no more than general verdicts. The problem lies in the idea that ready-made contents could be received by us independently of our acts of thinking. In fact, one could distinguish three different kinds of Given: 1) atomistic and nonconceptual, which is the variety endorsed by most non-Quinean version of classical empiricism, 2) atomistic and conceptual, which would be implicit in a re-enchanted nature and 3) nonatomistic and conceptual, as we could find in Quine. It seems that some passivity (or the idea of some receptivity) is crucially at odds with holism.
[xv] An identity theory of truth such as that defended by Hornsby (1997) could match finely with such an alternative: facts are the same as true thoughts.
[xvi] We present this second alternative mainly because we had believed for a while that it offered a promising way to conciliate holism and responsiveness to the world. We hope someone else could learn with our mistakes.
[xvii] The appeal to truth-makers in an account of truth is a move that often commits one to a conception of truth as an adequatio intellectus ad rem where there is an intelligible element of res that the intellectus has to respond to in order to reach truths. Confrontation between truth-makers and our beliefs as such, however, can hardly be rendered intelliugible (as Davidson 1983, 1990, 2001 has argued).
[xviii] McDowell (1994: II) argues that there is no priority between thinkables and thoughts––thinkables that are part of the world were not made prior to our thinking in any interesting way.
[xix] This can be developped in the direction of an account of a priori knowledge according to which nothing specific is to be known a priori but something has to be known a priori. Further, one could claim that some (synthetic) a priori judgments are necessary but we run into difficulties of different sorts if we try and specify what would be those judgments.
[xx] Knowable contents contrast with thinkable contents. We believe knowable contents, or even ‘knowables’, contrast positively with ‘thinkables’ (see Hornsby 1997, inspired by McDowell and Wittgenstein`s Tractatus) and ‘claimables’ (as in Brandom 1999). A knowable content has in itself an element of the world as it is a genuine object of knowledge.