a readable version

Thought Holism meets World Holism
Hilan Bensusan & Manuel de Pinedo

1. Holisms

Thought Holism (TH) is the thesis according to which thoughts are interconnected by their semantic features. This interconnection extends to beliefs, meanings and mental contents generally. Intelligibility of thought, and its relation to the rest of the world, ties together different thoughts in a larger pool. According to TH, thoughts in isolation cannot be recognized as such – TH is the opposite of the Thought Atomism to be found, for example, in many versions of a representational account of the mind.

World Holism (WH), by contrast, is a thesis about the world. It claims that parts of the world are interconnected by their modal features. These parts can only be described by means of the dispositional properties that connect them together in a larger pool. WH is opposed to various forms of World Atomism that take the world to be composed of modally independent parts.

In this work we begin to explore the relations between TH and WH. We do that by means of a main contention: that WH is the best metaphysical companion for TH: TH is both best understood and best grounded if associated to WH. This contention has implications for the placement of thought within the world: we claim that in a WH scenario, thought could be fully integrated with the rest of the world. There could still be ways for TH to be accommodated in a more atomistic metaphysical outlook, but we will offer reasons to pair it with a full-blooded holistic conception of the world – and suggest that TH at the very least strongly encourages WH. (At least if we consider the versions of TH and WH that we favour.) If we are right, what emerges is a fully holistic conception of thought and world with interesting consequences both for the place of powers, intentionality and normativity in nature and for knowledge, truth and the contact between thought and the rest of the world.

2. Elaborating versions of TH and WH

We take the most satisfactory version of TH to be close to that recommended by Davidson (1974, 1986, 1991, 2001) for a number of reasons: (a) it is a thoroughly non-empiricist version of holism about mental content, (b) it is based on the rejection of the first two dogmas of empiricism denounced by Quine, together with the rejection of a third dogma that would make sure that no empiricism is left (1951), (c) Davidson’s holism is also committed to an externalist account of mental content: the world affects the pool of our (mostly true) beliefs and (d) central to his endeavour to flesh out and defend a TH, there was the effort to indicate how a collection of thoughts can respond to the world. We begin by contrasting his own holistic account of this contact with a standard view, akin to empiricism, that would take thought to be responding to the world through some privileged points of access.

The standard, atomistic ways to present how thought can respond to the world often assumes that thinking could be intelligible while being utterly indifferent to the world. The view calls, then, for some part of the world to be more readily available to thought – what we call a bottleneck, a channel through which the world affects thinking. Bottlenecks can appear as sense data, immediately perceptible objects, passively acquired (non-conceptual) contents and various other forms. Often, empirical contents are taken to act as bottlenecks. Singular non-modal items get in touch with thought through the bridge of the senses. Empirical contents, also, could be conceptually loaded and still act as bottlenecks.[1] The picture has therefore two conjoined components: the appeal to a passive, receptive element in our thought and an atomism about our responsiveness to the world.[2]

Davidson’s holistic view, on the other hand, postulates no bottlenecks. The idea is that internal properties of thoughts, chiefly their interpretability, ensure that the world is reachable. The manoeuvre, as envisaged by Davidson (1974, 1986, 1987, 1991), tied thought and the world through truth and intelligibility. Consider the following modal skeptical challenge:

MSC.

Premise: (It is intelligible that) each of my beliefs can be false

Conclusion: Therefore, (it is intelligible that) all of my beliefs can be false

One can accept the inference and attempt to show that the premise is false. Such a path leads quickly to the postulation of bottlenecks: through thoughts that cannot be wrong I make contact with the world.[3] Davidson’s holist take, in contrast, would be to show that MSC contains no valid inference.[4] If not every belief can stand or fall on its own, we can begin to see how each of my beliefs could be false and still intelligible while that could not happen to all of them. A crucial ingredient of the argument is that we cannot establish a separation between that which fixes meaning and other beliefs – that my beliefs are to be assigned a meaning (an interpretation) under the light of other beliefs. That there is no separate pool of meaning-fixing beliefs is a consequence of the Quinean rejection of the first dogma. The rejection of the second dogma entails that no verdict from the world can impinge on less than a collection of interconnected beliefs. Confrontation with the world requires always a background of true beliefs in order for the verdict to be intelligible. Davidson’s holism deals in critical masses of thought: within a critical mass there is intelligibility, truth and contact with the world. Little can be said about each belief constituting the critical mass. However, reassurance that thought is not indifferent to the world comes from the critical masses that make each belief understood.

We can try to formulate the argument that supports the claim that some (sufficiently large) critical masses of thought enjoy contact with the world in terms of semantically interdependent beliefs (sib for short): a collection of n beliefs so that there is an m (1 < m ≤ n) so that if m beliefs are false the collection is unintelligible. A sib is a critical mass of beliefs and, for our purposes now, we shall consider sibs such as that composed by all of someone’s beliefs (call them a-sibs). The argument, that we called Holistic Contact between Thought and World (HCTW), could be then presented as premises (1)-(5) leading to the conclusion (6):

HCTW.

(1) All beliefs are equally about the world: no belief is intelligible purely in terms of its contribution to the interpretation of other beliefs (From the rejection of the 1^st dogma)

(2) Conversely, each belief depends on other beliefs to be understood and to receive a verdict from the world – understanding and confrontation with the world apply to critical masses of beliefs, to sibs. (From the rejection of the 2^nd dogma)

(3) Empirical content comes in the form of beliefs. (From the rejection of the 3^rd dogma)

(4) If any sib contained too many falsities, it becomes unintelligible and cannot be confronted with the world. (Definition of sib)

(5) If an a-sib were unintelligible (if it contained too many falsities), its falsity would become unintelligible – and talk of truth (within the a-sib) would become itself unintelligible. (From 1-4)

(6) There should be some truths in an a-sib.

If the argument is sound, there should be some truths within any intelligible sib.[5] Certainly, we lack the means to locate the truths within a sib; in fact, from a holist point of view, the truth of a belief always depends on others as none is self-standing. Furthermore, if the premise of MSC is true, we seem to need a measure of indeterminacy as to where the truths are located in a sib.

The version of HT we favour is broadly Davidsonian in that it is committed to HCTW and to the link between intelligibility and truth. We take the features of our version of HT to be a consequence of this broadly Davidsonian position. The most salient are:

α. A thought cannot be understood or confronted with the world on its own, it needs an environment of thoughts.

β. A thought cannot be understood unless we consider the effect it could have over other thoughts (for instance, in terms of the inferences it entitles).

γ. We can highlight a particular connection between thoughts only by blocking the effect of surrounding thoughts; it is only by holding something fixed that we can pinpoint them.

In contrast, holism about the world (WH) appeals to modal connections between its parts – they are integrated through these connections as each part carries dispositions. The drive for holism comes from a rejection of the claim that all dispositional properties are grounded in categorical ones. It follows that there are at least some grounding powers in the world. Dispositions (and powers) are somehow directed towards another part of the world – their identities require something outside them, they enjoy what Molnar (2003) called physical intentionality. Our preferred version is one according to which there is no purely categorical properties – either all properties are dispositional (Shoemaker 1980) or properties are both categorical and dispositional (Martin & Heil XXXX). According to a power-based WH, the world is full of modal connections and we cannot understand any part of it without making reference to dispositional properties. Powers affect and are affected and no singular item can exist if it is not in some dispositional relation with others. We tend to view powers as singular, dispositional tropes: each disposition is unique in its potential relation to other items. Physical intentionality depicts singular elements even though we often use general terms to describe powers (such as fragility, solubility, edibility). Powers are therefore utterly relational and therefore holistic but, at the same time, singular. The singularity of powers entails nothing concerning quidditates: they are discernible one from another only by means of their potential effects. Finally, and contrary to a tradition in the metaphysics of dispositions (that revolves around the work of Armstrong) we prefer to consider powers as affordances and not as properties because to our ears affordances evoke capacities in a stronger manner. [6]

WH contrasts with a mosaic-like Humean ontology where parts are put together contingently. In a power-based WH, to individuate connections between two items of the world, we need to postulate that the surrounding powers are kept fixed. In order to, say, predict something based on a conditional such as “if this salt is put in water it will dissolve” we need to assume that other powers are not affecting the salt. Such assumptions enable nomological statements to be predictive – and explain their failures; we can individuate connections between parts of the world and without them, prediction is impossible as each item of the world is subject to an indefinite number of powers. Certainly, while a mosaic-like world would make knowledge impossible, in a world where nothing can be isolated from the rest and everything affects everything else knowledge is equally impossible: knowledge needs graspable connections. To some extent, a device to introduce ceteris paribus scenarios is the only way we can focus on particular connections. They introduce an element of modal inanimateness capable of separating a particular connection from the others. Without them, dispositional properties intertwine all parts of the world and everything can affect everything.

The features of our favourite version of WH are consequences of the power-based ontology above. The most salient are:

α. A part of the world cannot exist on its own, as it needs other things to affect and be affected by it.

β. A part of the world cannot be brought to focus unless we consider the effect it could have over other parts (for instance, in terms of what it causes).

γ. We can highlight a particular connection between parts of the world only by blocking the effect of surrounding parts of the world; it is only by holding something fixed that we can pinpoint them.

3. Some reasons to couple TH with WH

Having presented the features of our favourite versions of TH and WH we can now move on to present reasons in favour of the coupling of both doctrines. These reasons range from the advantages of a consistent rejection of atomism to considerations related to the integration of thought to the world. They are part of the picture of the thorough holism we are putting forward.

3.1. An atomist world is epistemologically and semantically implausible

A modally poor metaphysics entails various forms of scepticism, chiefly related to problems of induction. A modally poor world can only relate to thought by means of bottlenecks: a world of detached pieces is to be accessed by thought in a piecemeal way where whatever that is sensed works as the starting point. Thought finds no dispositional connections to exploit, the ties between affordances and what they physically intend becomes no more than a projection of our interests or habits and no longer part of the ontology – they are at most the product of our sovereignty. Humean accounts of modal connections encourage the idea that dispositional affordances cannot be part of the world and are exiled to a second creation, that responds to nothing but human convenience. An atomist and modal-poor ontology entails almost immediately a measure of scepticism: the world cannot be called to warrant any projection from what is received through the bottleneck towards what is still unobserved. Holism concerning thought rejects this projection legacy by showing that there is no part of thought where the world cannot be involved. TH has then exorcized the epistemological consequences of world atomism – and that strongly suggests that another metaphysics is in order.

We have said that in order to state a specific dispositional connection, we have to assume ceteris paribus scenarios. One can then say that this assumption introduces a measure of our sovereignty. In fact, a suspicion can arise that WH does no more than reverse the Humean image of a modally poor world associated with a second creation of dispositional connections – the world become a pool of powers where we select some assuming the rest is going to be kept fixed. It could be as if we need fixity and constancy in order to attain knowledge of the world and that cannot be provided by the world itself – as Hume held that modal affordances cannot be provided by the world itself. We will have more to say about this suspicion later but two related points of dissimilitude between the two images deserve attention now. First, what we do when we postulate a ceteris paribus scenario in order to act or think is to subtract something from the world and not to add anything. Our intervention would be just to establish a focus, to depict a part of the world disconnecting it (modally) from the rest. The raw material of modal connections is already there and is itself no product of our intervention. The contribution of thought is no more than selecting a part of the world to focus on.[7] Second, the postulation of a ceteris paribus scenario is the effect of a power – that thought, but not only thought, entertains. Knowledge in fact requires both modal connections and fixity, and the world is such that some interventions can produce greater fixity. Such a postulation is not something alien to an ontology of powers – it is not something added to it from outside. The manoeuvre is different from that of inserting modal connections in a world of distinct individuals in that the products of thinking while selecting powers could be accounted for in terms of what the world is made of – mind doesn’t present itself as alien and detached from the world it intends.

Humean ontology favours separate and modally inanimate items that establish only contingent connections between them. This is semantically implausible because we need stronger connections between items – normally necessary connections – in order to ensure connections between different predicates. Humeans hold that a is necessarily connected to b only if a is identical with b and therefore (in most cases at least) causation cannot be a necessary connection. So, there is a necessary connection between chordates and creatures with a heart (and a causal tie between chordates and creatures with a kidney). The Humean world is not modally inanimate, it is analytically animate. Necessary connections are reached through the trick of making some connections available to reason alone. If, according to (1) in HCTW above, there are no beliefs that are solely a device to understand other beliefs, then there could be no belief in a necessary connection that is provided by reason alone. In other words, Quine’s rejection of the first dogma can cut deep against Humean assumptions about analyticity that invokes it to provide the concealed necessities needed. Whenever we have observed occurrences of sentences where expressions like ‘chordate’ appear, we have also observed we were prone to accept sentences where ‘creatures with a heart’ appear – as we take both expressions to point to the same part of the world. Any further inductive step is unwarranted. Likewise, we have observed occurrences of sentences where expressions like ‘chordate’ appear, we have also observed we were prone to accept sentences where ‘creatures with a kidney’ appear – as we take both expressions to point to the same part of the world. Here again, any further inductive step is unwarranted. Humeans would like to say that a further inductive step is available in the first case – as it is a matter of reason that there should always be identity between some expressions, a matter of reason that relies solely on truths by virtue of the meaning of the expressions. Humeans hope to get necessity out of semantics – as a by-product of expressions having meanings. Quine’s rejection of the dogma suggests that there is no such shortcut: modal connections have to come from the world or will be no more than a projection from our habits. Once TH is adopted, there is no exception for analytical connections.

The reasons to move towards TH leave epistemological and semantic discomfort with an atomist, Humean image of the world. TH begins to appear more at home in an ontology of powers.

3.2 Holism, predictive truths and modality

A central element of the atomistic, mosaic-like view of the world is what Molnar (2003: 181) calls ‘Humean distinctness’. The idea is that (apart from the appeal often made by Humeans to analytical, de dicto necessary connections between items) items are only connected to each other contingently. Such a thesis, that there is no modal glue among items in the world, can be shown to be false by an extension of HCTW above. The extended argument purports to show that MSC above is an invalid reasoning because doubting cannot be done in an atomistic, cumulative manner. Specific doubts, the argument shows, require specific grounds for knowledge, like thought, cannot rest on atoms.

The extension of HCTW is done by considering other sibs. We can take some subsets of what we understood to be an a-sib – the critical mass formed by all of someone’s beliefs – to be sibs themselves. Consider the collection composed by all beliefs that make direct or indirect reference to what is not yet observed (not only ‘the sun will rise tomorrow’, ‘the clock will carry on ticking in the next minute’, ‘the rooster will sing in the morning’, ‘the water will boil in 10 minutes’ but also ‘the next emerald to be found will be green’ or ‘the next fossil to be found here will be of the same species’). The collection of beliefs about what is not yet observed – call it the collection of predictive beliefs, P – is a sib for if we take out enough beliefs from it (or show enough beliefs to be false), the remaining ones are going to be unintelligible. A predictive belief can only be false if a number of other predictive beliefs are true – otherwise falsehood about predictions ceases to make sense. We can call a sib a central sib (c-sib) for an a-sib if the following condition is met:

(7) If all beliefs in a c-sib are false, no other belief in the a-sib can be true.

We take that there are some c-sibs for any a-sib as there are some general features of beliefs and truth that require that some beliefs be present in any a-sib (even though no particular belief has to be present in any a-sib). Further, we claim that:

(8) There is at least one c-sib for any a-sib for P is a c-sib for any a-sib.

We take (8) to be a reasonable consequence of the centrality of prediction for belief. Some predictive beliefs have to be true in order for predictions to make sense and beliefs to extend beyond what has been observed. Beliefs are intelligible only against the background of other beliefs that link what has been observed and what is expected to be kept fixed – beliefs that establish a natural clock against which predictions about the future can be understood and put to test. Without those beliefs, there is no sense in using future observations to confront beliefs as ‘future’ can only be understood against the background of beliefs about natural clocks.

If we add (7) and (8) to HCTW, we have all the premises of the extended HCTW (EHCTW) that, from (1)-(8), would conclude that:

(9) There should be some truths in aP .

It follows that we can only question whether the sun will rise tomorrow against the background of accepted predictions concerning tomorrow – predictions that we tend to place in the very definition of ‘tomorrow’, such as that the rooster will sing or the clock will be indicating midnight.[8] Clearly, these predictions are as much part of P as the other predictions. Furthermore, some of the beliefs in the collection have to be true, otherwise it is impossible to recognize thought within it – and, therefore, it would be impossible to spot failures of prediction. Hume assumed that if the future resembles the past this would be a matter of fact and not a question that could be established by reason (alone or aided by experience). If all of our predictions were wrong, however, their falsity would disappear as they would cease to make sense. P, as a collection of inductions, is intelligible only in so far as some of its members are true – all of our inductions cannot be false. If the argument holds, the truth of some inductions is established.[9]

If some predictions are right, there are some modal connections between what has been observed and what has not – predictions exploit the modal ties and typically assume that in some cases, future observations will resemble past observations. That some of our beliefs about future observations are true show that our true beliefs cannot be concentrated on some observed items of the world while having no implications for the others. Notice that it is Humean distinctness and modal scepticism that are challenged by the argument and the appeal to future observations (and the future) is not central. In a four-dimensionalist ontology (XXXX) we can take predictions to be ties between different parts of space-time – the argument establishes that some modal tie has to be in place. EHCTW is an extension of HCTW, an argument that is part of our preferred version of TH. If we further hold (8), which is a reasonable feature of classes of beliefs forming a-sibs, we conclude that some modal ties between the items of the world have to be in place. TH leads us to some sort of modal anti-scepticism that further fuels WH.

3.3 Thought is part of the world

We can find support for WH if we manage to see the arguments for TH as a special case of a broader holism. Physical intentionality could be a starting point to consider thought as part of the world. Molnar (2003: 60-66) argues convincingly that the four features normally associated with Brentano’s account of intentionality can be only slightly modified to characterize an item with physical intentionality: i. it is directed to something beyond itself, ii. it could be directed towards something nonexistent, iii. it is disposed towards exemplars and prototypes rather than specific items and iv. it is sensitive to the way the intended item is presented. (We mean by physically intentional what satisfies this Brentano-Molnar characterization.[10]) Thoughts have a dimension of physical intentionality. Thought contents are directed towards other contents centrally because they are dispositional ingredients to produce other thought contents. Here it is interesting to consider the debate between inferentialists and representationalists – the latter being the ones who consider contents as categorical properties that carry a representation inside them quite independently of other surrounding thoughts and the former being those that consider contents in terms of their power to infer more contents. The debate is many-faced and the two positions have as many versions as they are intermediate proposals recommended (see, for example, Brandom 1994…). However, TH tends to side against the representationalist conception that a thought could be fully characterized in terms of a fixed states determined by what it is about. Inferences, to be sure, are not the only things that are in the powers of a thought: a thought can affect action, perception and several other abilities that require conceptual capacities (even though those capacities surely could be understood broadly in terms of inferences, see Brandom, 1994). The stress on inference, nevertheless, is enough to bring to light a feature of intentionality that is present in physical intentionality: the affordances of an item have to do with its capacity to affect other items – and we can think of inferentiability as a power. To think of a thought content in terms of its inferentiabilities contrasts with the image where it is first seen in terms of what it is about. Our insistence on inferentiabilities as powers doesn’t entail that a thought content has to be taken as a bundle of inferentiabilities (of powers) but what it is about cannot be brought to view without the bundle. Thoughts are in dispositional relation one to the other and truth is attained only within the scope of critical masses (sibs) where contents inter-animate each other. The elements in common with physical intentionality can be brought to view in the following (a)-(b) to (c) inference:

(a) Snow is white [a belief within an a-sib]

(b) There is snow in the landscape [acquired through perceptual capacities, or by testimony]

(c) There is something white in the landscape

Here, i. we understand (a) in terms of its power (among others) to infer (c) on the presence of (b). ii. In the absence of (b), (a) would still be directed towards (c) – along with other powers it has – while not being part of the inference (actuality of unmanifested powers). iii. Surely, (b) is not more than a prototype – it can be presented as, for example:

(b’) Peter saw snow falling last night [by testimony]

(b’’) There is always snow in this area at this time [a belief]

(b’’’) I can see the snow in front of me [by perception]

Finally, iv. one has to recognize the snow in order to conclude, from (a), something like (c); which is not possible if snow is presented in a very different format (for example, filling pink pills).

This is not to say that there is nothing specific about the powers within thought. In fact, mental intentionality involves some special connection to rule following through which they are affected by, and acquire the powers to affect, the thoughts of other people. Full-fledged thoughts are interpretable and interpretation makes them recognizable as thought. We acquire new powers when we acquire the all-entwined capacities to use and recognize concepts – conceptual inculcation affects our affordances, our capacities to be active. Concepts make us respond to the world in ways that contrasts with our inclinations and provide an environment rendered fixed by assuming that some powers are not present. So, the concept of red is acquired together with the assumption that often the light conditions are going to remain the same – and no further power will intrude this artificially limited set-up. Concepts introduce lab-like limitations in our environment and this is part of our activity of focusing on parts of the integrated whole that compose the world: the contribution of our sovereignty is to fix some modal connections in the world by rendering some further powers sterile. The introduction of fixity – of ceteris paribus scenarios – is arguably not an exclusivity of conceptual activity (and of our sovereignty over our world-view). Many physical and biological powers depend on a fixed environment: the frog (physically) intends a fly and is not geared to an environment full of fly replicas; salt will not disolved under extreme temperatures, the bee goes for a flower-looking object and not towards a flower pill that might contain everything it needs, etc. Features iii. and iv. of the Brentano-Molnar characterization of physical intentionality suggest that other powers have the capacity to produce a ceteris paribus scenario around them: iii. the capacity to be directed towards exemplars and prototypes of an item and assume its specifics have irrelevant powers and iv. the capacity to be directed to something in a specific mode of presentation assuming that other powers could not affect the presentation of the directed item. We call these devices that act on the assumption that some powers are indifferent to others ceteris paribus devices (CPD for short). Thought – and arguably mental intentionality – is a CPD but we claim that they are not the only ones. CPDs are grounded on physical intentionality. Our capacity to project fixity in the world – in contrast with the capacity to project modality postulated by Humeans – is shared by other CPDs in the world. Our own projection can then be understood on the light of CPDs in general that, in its turn, can be understood in terms of physical intentionality.

TH can be viewed as a response to the failed attempts to establish a powerless (or inanimate, or nondispositional) realm within thought – like a bottleneck conceived as a blank slate to be impressed by the world. Such a slate will be a powerless effect of the perceived powers around us. Holism claims that there is no merely passive part of thought – powers are everywhere. We can select some powers and detach a part of the world but that is the work of a CPD acting locally – nothing is in itself powerless. If it is plausible that thought is part of the world, we should consider what counts in favour of TH as a special case of what counts for WH. As an exercise, we can attempt to expand on the three first premises of HCTW above (the rejection of the three dogmas) so that they are formulated as general claims about world holism:

(I) There is no atomic, fixed necessary connection that could be a truth-maker for our discourses on necessity. A modal connection is never unaffected by the other surrounding modal connections. There is no separate realm of necessary connections enjoying a special status that would make them effective comes what may – no separate realm of analytical connections, no separate realm of nomological connections. Therefore, no nomological necessity could be isolated from all the other powers affecting each other. There is no principled distinction between fixed nomological necessity on the one hand and ordinary powers – only as an approximation we can take laws as fixed connections (typically, through ceteris paribus clauses).

(II) As a consequence of (I), there is no picture of isolated parts of the world that could be taken as more than approximations. Our ontology should include all the intervening powers. Descriptions of thought and of the world should deal in critical masses.

(III) No item could be isolated from its powers (potentialities, capacities, conceptual schemes for thought). Therefore, there is no underlying substrata to objects and no underlying quidditas to affordances that can be detached from its powers. Thought content cannot be separated from its conceptual (for example, inferential) capacities and no part of the world could be separated from its powers.

The rejection of all kinds of fixed necessary connections (claim (I)) follows from the all intruding character of powers – they are holistically connected to whatever come their way and they are not limited to a fixed, easily detectable scope. This puts together the rejection of semantically necessary connections and eliminativism about natural laws as recommended by Mumford (2004): in both cases we assume no fixed necessary connections that are immune from other surrounding powers. The emerging picture is one where powers compose neither a mosaic nor a jigsaw with fixed positions but rather a set of moving magnets where attraction and repulsion comes from all parts and no connection is stable. Fixed necessary connections are not truth-makers, they are just part of our worldview, consequences of a CPD. We need some ceteris paribus clauses to provide fixity in order to focus on specific powers.

Claims (II) and (III) summarize the analysis of thought in terms of powers – there are no isolated parts of the world and there is no individual part that can be separated from its distinctive powers. They show further how TH can be seen as a special case of WH: conceptual capacities and physical dispositions seen as powers. Thought, as the rest of the world, is bounded by no fixed necessity and is not composed by individual parts that can be dissociated from its powers. Further, thought builds on dispositional affordances to produce a CPD that is, as any other, capable of introducing fixity and detachable parts in the world. If all these elements sketch a convincing image, it is an image of a thorough holism where the interdependence of thought contents is further illuminated by an ontology of interrelated powers.

3.4 Thought as a special part of the world

Last section presented the integration of thought and world through powers and physical intentionality as a reason to couple TH and WH. There are, nevertheless, strong intuitions that some features of thought – chiefly to do with its normativity, its rule-governing character and its connection to (reflective, conceptual) knowledge – make it special with respect to other parts of the world. These intuitions stress that there is a difference between mere physical intentional states and states that are about something else. In this section we will try to suggest that making room for these intuitions could provide more (and not less) reasons to embrace both TH and WH.

We have argued that powers could take us all the way from a TH to WH. In other words, once TH is embraced – at least in the version we favoured which is fairly Davidsonian and in any case committed to a degree of semantic externalism – holism seems to be hard to be keep away from ontology for the reasons above. There is nonetheless the temptation to tame some of the (world) holistic consequences of TH by embracing one form or another of partially atomistic metaphysics chiefly by softening claims I-III above in order to accommodate the supposed special character of thought. These temptations appear when TH is held together with a conception of nature where laws and determinate necessities underlie thought and make it look anomalous and, in some cases, unnatural. It could seem odd that a world of natural necessity would ground rule-based processes and that fixed necessary (nomological) connections were the elements available to account for our capacity to respond to reasons. An image of nature as a realm of interconnected laws can make it difficult to present normativity as part of it; a difficulty that stands out not only in naturalist projects to place thought in a natural scenario but also elsewhere in attempts to make place in nature for norm-governing behaviour.[11] The underlying image of nature – and the underlying ontology – refrain the holism about thought to spread to the world. It is this underlying image of nature that furthers the impression that thought, and mental states in general, are somehow different and anomalous.

Davidson appears sometimes to be prey of trying to put together TH with an atomistic take on natural necessity. He is at once the most profound champion of epistemological and semantic holism and a defender of the need for laws to make sense of causal connections. Thought ends up having to be anomalous and we can see a gulf emerging between the parts of the world capable to think and anything else. Further, the gulf is between mind and world as the former is thought of as revolving around contentful, normative states where the latter is made of categorical properties and laws. Davidson postulates that thoughts are causally connected to the world and embraces a Humean view of causality according to which there should be some description of thought where a law holds between thought and what causes it. This appeal to the need for laws under some description – that McDowell (1985) dubbed the fourth dogma of empiricism to be abandoned when the third is dismissed – introduces fixed necessity into the picture of thought. There is a way, then, at least in principle, to recognize thought without interpreting it – by exploiting the nomological connection. The appeal to fixed nomological necessities makes thought grounded on laws (and the categorical properties causally connected by them). Of course, at least in the best interpretations[12], thought content depends always on other contents and no ground can determine what a thought is about. Thought enjoys a degree of sovereignty but only to the extent that it becomes anomalous – detached from the rest of the world, subject to no (psychophysical) laws. Further, it is causally connected to the rest of the world through some grounding: Davidson’s so called weakly Humean account of causality can be read as a way to put forward the thesis that there should be a categorical property behind any thought that connect it to its content. By being coupled with WH, Davidson’s TH would be freed of the idea of anomaly of thought and thought could be conceived out of materials that are themselves dispositional affordances with physical intentionality. No categorical property would be postulated to ground thoughts – and no laws will be needed to assure a causal connection between a thought and its content. The connection would be just a matter of powers.

Powers are behind both physical dispositions and conceptual capacities. Surely, a crucial element for thought is related to its capacity to be sensitive to rules, to follow them and to respond to concepts: thought is not only (physically) intentional but it is normative. There is a gap to be breached between physical affordances and normativity but the latter might just start to look less unnatural if we take the former to be physically intentional. The Brentano-Molnar characterization of physical intentionality (especially related to iii and iv above) can help viewing the intentionality of thought in a continuum with physical powers (and thought in line with other CPD devices). So, for instance, a bee seeks a flower no matter its shape or colour – as we conceive of a man independently of his height. A bee is not happy with every presentation of a flower, a pill could contain all that it needs but it wouldn’t go for it. Assuming a bee has no thought contents, it follows a physical intentionality out of its inclinations. No matter how convinced we are by Millikan’s (XXXX) attempts to understand intentionality in terms of proper functions, we can draw the conclusion that inclinations are themselves intentional. In fact, if intentionality can be taken to be present already within inclinations – concept acquisition introduces normativity in a picture that is already intentional. The pupil of the famous example of Wittgenstein’s Investigations (PU, I, §185) can be said to have a particular intentional inclination at each stage of the process of learning how to follow the rule “+2”. The pupil’s behaviour can be seen as a product of a systematic error due to having seen the examples given as cases of something like “+4 when the sequence is greater than 1000”. This can be described as an intentional inclination of the pupil. Wittgenstein acknowledges that there could be fully articulated (mistaken) intentional items by saying that there is no way to distinguish a systematic from a non-systematic error in a pupil. Inclinations don’t have to be taken as randomly produced, they could be as intentional as the rule that in the expected end of the process the pupil will supposedly entertain. The important difference – between inclination-based and rule-governed action – is then not a question of intentionality, as both rules and inclinations satisfy the Brentano-Molnar conditions for physical intentionality. The difference is one of normativity – this is what is inculcated in the process of learning to follow a rule. (Physical) intentionality was there from the beginning.

Normativity is part of the specific intentionality of thought. Naturalizing normativity would be the name of the project to understand it in terms of physical intentionality – not necessarily to explain it away in terms of physical intentionality. Normativity has a special way of dealing with iii and iv: concepts deal with prototypes and exemplars and create opacity in language because they are relative to their mode of presentation (they introduce non-extensional contexts)[13]. A normative, conceptual CPD builds a conceptual interface with the world that carries within in some ceteris paribus assumptions. These assumptions are typically in the form of semantic connections (or so-called analyticity). This conceptual interface creates a scenario for intentional items where some powers are considered while others dismissed. Conceptual interfaces, like CPDs generally, are also powers, they are capable of interaction with other powers. Because of that, they can bring new things to the world. That their intervention always involves keeping some powers fixed, inanimating them, so to speak, should not be an obstacle for their creative nature. Forcing some powers into inertness is their way to bring to the fore other powers, to make them salient, to increase the possibility of interacting and combining with them into new powers. For instance, the move from being inclined to eat whatever it has a certain appearance to mastering the concept of food increases fixity and inanimation (by ignoring all sorts of other powers that could affect edibility) and at the same time affords you to eat a broader variety of things.

Inclinations, on the other hand, are powers but are not available as such to thought – they are directed towards some items without being about them, without having (non-conceptual) content. Leaving aside the epistemological and semantic details of such a view of inclinations, we would like just to point out that concepts bring up normativity through its special resources to create fixity. Those resources, we believe, can be fully understood in terms of powers. In this (perhaps restricted) sense, normativity can be naturalized.

We can also understand knowledge – and its ties with thought – better if we put together TH and WH. TH teaches us that true beliefs has to be present in any critical mass of thoughts. It is reasonable to take knowledge as having to be present always in thought.[14] In fact, we can think that knowledge is one of the powers of thought and it is easy to understand it in terms of dispositional affordances: we can afford to act and think things because of what we know. Surely, reflective, conceptual knowledge requires special abilities (powers) that are inculcated in us through our upbringing within language; the constituents of these abilities, however, can be understood, as we saw, in terms of powers. Knowledge, like any dispositional affordance, is holistic – it acts depending on its surroundings. Holistic knowledge is such that we cannot point at where exactly it lies – it’s elusive. It depends on a critical mass of thought and we cannot pinpoint where, within that mass, knowledge lies. TH gives us means to ensure ourselves that we know about the world – but we have no means to determine what exactly we know about it. The picture of WH is one of interconnectedness through modal ties. Reflective, conceptual knowledge exploits these ties and makes further powers available for thought and action. The search for knowledge lies in a framework of concepts and norms but it’s also akin to that of other CPDs with physical intentionality. Our knowledge, nevertheless, can be directed to things as far as our concepts can reach. Directedness is extended by conceptual means. The holism of thought introduces, therefore, a further holistic dimension to knowledge: our knowledge responds to everything our concepts can reach, it acts over a greater scenario than other CPDs. Knowledge (and concepts, rules, normativity) makes thought special – but only in the sense that it is a special power and, as such, a special part of the world.

4. Future work

This work is meant to be programmatic in nature. It starts considering the connections between TH and WH and arguing that there are reasons to embrace both. In this sense, it doesn’t intend to do more than begin to explore the connections and much work needs to be done to highlight the consequences of such a coupling and to further develop some of the ones discussed in this paper.

We would like to briefly mention a few. First, the application of an understanding of knowledge as power to bring new light to some traditional puzzles in epistemology, such as those related to the definibility of knowledge by means of the idea that knowing something involves powers absent in mere true belief. Second, the idea that a conception of thought as power, by stressing that thoughts involves acts of thinking no less than articulated sets of contents, could open new ways to think about the debate concerning non-conceptual content. Third, more work needs to be done to elucidate the role of CPDs in what distinguishes actuality from mere power, which also relates to the debates about categorical properties within an ontology of powers. Fourth, the relation between powers and truth seems to be fertile: thoughts are truth bearers while other powers are somehow truth-makers – is truth a relation between powers? Fifth, the role of thought within WT deserves some consideration: if thoughts are powers amongst powers, they are subject to interaction with other powers and, in that sense, equally constrained by them; any temptation to view thought as capable of embracing reality as a whole would then subside.

We believe there are interesting philosophical results in all of those avenues. Further, we trust to have provided some prima facie reasons to add to an ontology of powers to a thoroughly holistic account of thought.

References unfinished

Bensusan, H & M. Pinedo (2007)

Brandom, R. (1994), Making it Explicit, Cambridge, Mass.: Harvard University Press.

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. (1986a), ‘A nice derangement of epitaphs’

Davidson, D. (1986), ‘A coherence theory of truth and knowledge’, in Subjective, Intersubjective, Objective, Oxford: Oxford University Press, 2001, pp. 137-153.

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. (2001) ‘Externalisms’, in Kotakto, P., Pagin, P & Segal, G. (eds.) Interpreting Davidson. Stanford: CSLI, pp. 1-16.

Ellis, B. (2002) The Philosophy of Nature: a Guide to New Essentialism, Chesham: Acumen.

Hawthorne

Martin, C. & J. Heil

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. (1985), ‘Functionalism and anomalous monism’ xxx

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.

Millikan, R. (xxxx)

Molnar, G. (2003) Powers: a Study in Metaphysics, Oxford: Oxford University Press.

Mumford, S. (2004) Laws in Nature, London: Routledge.

Nagel, T. (1999)

Pinedo, M. & H. Bensusan (2006)

Quine, W. v. O. (1951) ‘Two dogmas of empiricism’,

Rorty, R. (1986), ‘Pragmatism, Davidson, and truth’, in Lepore, E. (ed.) Truth and Interpretation, Oxford: Blackwell, pp. 333-355.

Shoemaker, S. (1980), ‘Properties, causation, and projectibility’, in Cohen, L.J. & Hesse, M. (eds.), Applications of Inductive Logic, Oxford: Oxford University Press, pp. 291-312.

Williamson, T. (2000), Knowledge and its Limits, Oxford: Oxford University Press.

Wittgenstein, L. (PU), Philosophical Investigations, Oxford: Blackwell; trans. of Philosophische Untersuchungen by G.E.M. Anscombe, edited by G.E.M. Anscombe and R. Rhees, 1967.

Zinkernagel, H. (xxxx)

---

[1] McDowell’s minimal empiricism (1994) comes to mind. McDowell’s bottlenecks are not committed to modal skepticism: any kind of content can be acquired by passive exercises of conceptual capacities. See Bensusan & Pinedo (2007).

[2] The bottleneck image carries its (atomistic) metaphysical burden. It is required that there are singular and modally inanimate items to be channelled through the bottleneck. These items fall into the bottleneck view with a measure of what is often called a Humean metaphysics (Ellis 2002, Mumford 2004, Hawthorne 2006).

[3] Foundationalism concerning empirical knowledge comes to mind. The search for foundations for empirical knowledge can be pursued through the postulation of bottlenecks. Not every foundation, however, can be seen as a bottleneck (if you take as the foundation whatever explains best our empirical knowledge). Conversely, not all sort of bottleneck images postulate a foundation for empirical knowledge (consider, for example, Brandom 2004 as discussed in Pinedo & Bensusan 2006). Accordingly, friends of bottlenecks do not necessarily take the premise of MSC to be false.

[4] The validity of modal inferences from premises each-premises to conclusions with all-conclusions is not easily decided by an appeal to a logical form. Consider the two following examples: (i) Each child in the playground can break a leg therefore all children in the playground can break a leg; (ii) Each player in the lottery can win therefore all players in the lottery can win. It is clear that while (i) is valid, (ii) is not. The holistic argument is that MSC is closer to (ii) than to (i).

[5] Surely, a sib is not just a collection (or a body) of beliefs but rather a network. Beliefs are semantically structured in a particular way within the sib. The sphere Quine mentions in the end of Two Dogmas is an example of structure within a sib.

[6] Our version of WH involves a realist account of affordances that differs, for example, from Molnar’s selective realism about properties in general (2003: 25-28). Molnar starts out considering that there should be no isomorphism between predicates and properties. We tend rather to understand that we are close to an isomorphism between concept-applications and affordances. We take affordances to demand all sorts of conceptual abilities for their specification – including family-resemblance concepts and complex predicates. Of course, not all predicate imaginable is in isomorphism with singular affordances, but we should limit ourselves to concept-tokens, application of concepts within acts of thought. We see concept-tokens as capable to attempt to depict singular affordances. Concept-types, on the other hand, play no central role in our characterization of affordances. Here we are in line with the picture of interpretation of language by passing theories in Davidson (1986a).

[7] Molnar’s proposal regarding physical intentionality paves the way for the intentionality of thought to ocupy the same plane as physical powers and to partially simplify the problem of the naturalization of intentionality. We will come back to this in section 3.4.

[8] The line of argument that guides the extension of HCTW has been suggested by Zinkernagel (xxxx) and McDowell (1995).

[9] In a sense, it can be said that this is established through reason in a way that is unaided by experience (and by any appeal to habit or to any Kantian transcendental reasoning). The holistic image of reason that emerges, however, would not allow for a separation between matters of fact and matters of reason.

[10] The characterization is in fact only Brentano’s, but Molnar made it, contrary to Brentano’s original purpose, a characterization of physical intentionality.

[11] We have maily McDowell (1994) in mind. His second-nature naturalism tries to put normativity within a broader framework of what is natural. He recommends that we view the realm of laws as part of nature – embraced by a broader space of reasons. The gap between the realm of laws and the space of reasons, however, seems large enough to make his reformed image of nature sounds like too extreme a departure from the more widely accepted image of nature he is criticizing (the one that includes only the realm of laws). By contrast, viewing nature as capable of physical intentionality brings together what underlies both the realm of laws and the space of reasons; the gap is not bridged but is clearly shortened.

[12] Sometimes Davidson is read as implying that a brain in a vat could only thought of the eletric wires as it is caused by them (McDowell, 1994: xx, Rorty 1986). Even though Davidson is very close to this at times, we believe that he would have better things to say about a brain in a vat (see Malpas, 2005).

[13] The discussion of opaque contexts typically (but not justifiedly if we take the Brentano-Molnar characterization to be about physical intentionality) revolves around the non-extensionality of some conceptual constructions (for example, in much of the discussion of Frege).

[14] The alternative cogito Thomas Nagel (1999) attributed to Davidson is a good way to present the idea: je pense donc je sais (I think therefore I know). If justification is to be needed for knowledge, a TH-like argument can show that we need to have some justification to begin with in order to have any justification at all. We, however, tend to embrace some form of primitivism concerning knowledge where justification itself cannot be understood without resorting to knowledge (see Williamson, 2000).