DISPOSITIONAL BELIEF, ASSENT, AND ACCEPTANCE

Pascal Engel

 

Dialectica, 53 , 3 / 4 , (1999) , 211-226

 

 

In a remarkable series of essays [1], Ruth Marcus has argued, against a number of contemporary language-centered theories of belief , that belief is not a relation to sentences or to linguistic entities. On her view, belief is a relation between an individual and a certain kind of object, which is neither a sentence nor an internal representation nor a proposition in the sense of an entity which is truth evaluable, but an actual or a possible state of affairs which obtains or not. It is a proposition in the Russellian sense of a complex entity (which may be non actual or possible) made up of actual individuals and relations. An important consequence of this view is that one cannot really believe, but only claim to believe, a contradiction, since a contradiction is an impossible state of affairs, and belief is a relation to a possible state of affairs. This automatically blocks such belief ascriptions as those that are familiar from Kripke’s puzzle[2], which seems to lead us to attribute to an individual a contradictory belief.  On the subjective side of the relation of believing, Marcus takes belief to be a disposition to act as if the corresponding state of affairs obtained. It is not a disposition to assent to a sentence or to a linguistic representation. This blocks the extended use of the familiar “disquotation” principle used by Kripke to frame his puzzle, according to which if a subject sincerely assents to P, then he believes that P. Divorcing believing from its usual manifestations in adult humans, such as assenting to a sentence, Marcus claims, allows us also to grant beliefs to non linguistic creatures, such as animals or infants, and to make room for unconscious or tacit beliefs, which, on linguistic accounts of believing, are not genuine beliefs or are not beliefs at all. It also permits a more adequate account of the rationality of belief, by tying rationality not to coherence of sentences but to a more general rationality of action.

      Although I do not want to commit myself to such entities as Russellian propositions or states of affairs, I think that the other features of belief that Marcus stresses are essential, and I agree with her that the linguistic accounts of belief are unable explain them. Here I shall not discuss directly all the aspects of her views about belief, but I shall try to give an analysis which bears a number of similarities with hers and supplements it in some respects. In particular, her view seems to me to be unable to account for another dimension of belief, which has close ties both to dispositions to act and to acts of assent, but which is reducible to neither of them: belief as acceptance, a state which has intentional and active components[3]. On the view presented here, belief is a generic attitude, which has most of the properties which Marcus emphasises, but it has also many subvarieties, which make it a layered and diverse, rather than a unitary, phenomenon.

 

I

 

         Marcus’s conception of belief is summarised in the following definition:

D. An agent believes that S just in case (1) under agent-centered circumstances such as desires, need, and other psychological states including other believings and (2) external circumstances (3) the agent will act as if S obtained, i.e will act in ways appropriate to S, where S is a state of affairs, actual or non-actual.[4]

 

This definition, which she adapts from Braithwaite, which himself adapted it from Ramsey[5], captures most of the interesting characteristics of a dispositionalist account. The mention of desires and other mental states such as believing avoids the familiar problem of reductive analyses of belief as a disposition to act: a belief is such a disposition only given other states such as desires, as well as other beliefs. Most contemporary functionalist views, which I take to be compatible with this one, admit that belief cannot be defined in a non circular way. For this reason, let us call it the “dispositional-functionalist” view. The view also incorporates Ramsey’s insight in “Truth and probability ”(1927): it says that given desires, belief is a disposition to act in ways which maximise the expected utility of the agent desires. The troublesome feature of (D) lies in the mention of states of affairs, possible or actual. There are familiar problems about  theories of belief content which take it to a be Russellian proposition made up of object and property.[6] But let us forget for the moment Russellian propositions, and let us try to work with an account which has some important similarities with Marcus’s.

        The account is Stalnaker’s (1984) analysis of propositions in terms of sets of possible worlds. Stalnaker’s first step (“the pragmatic picture”) consists in taking an individual belief believing that P to be a disposition to act in ways that would tend to satisfy one’s desires in a world in which P is true (Stalnaker 1984: 15). In other terms: an agent who believes that P has a behaviour of a kind that would maximise, other things being equal, the expected satisfaction of his desires, in a P-world, or to behave as if the actual world were a P-world. The problem with this first step is with the “would tend”, and “other things being equal”. In other terms, individual beliefs underdetermine behaviour. But beliefs do not come one by one,: they have an holistic nature. Stalnaker’s further step consists in identifying the content of the total  belief state of the agent with his dispositions to act: the content of his whole belief state is given by the set of worlds in which P is true, and there might be a number of individual beliefs in this state (1984: 82). As Stalnaker says: “If one conceives of beliefs in this way, they look like something negative: to believe that P is simply to be in a belief state which lacks any possible world in which P is false.” (1984 : 69) For example, suppose that a soldier believes that there are mines in this field. His behaviour is consistent with his desire to avoid walking on a mine, and with a whole set of beliefs – that there is a mine near the tree, that mines are dangerous, that the mine is going to be triggered if he steps here, etc. The possible world analysis does not cut finer. The agent believes that P if and only P is true in the set of P-worlds compatible with his belief, i.e in his belief worlds.

Stalnaker’s account does not identify propositions with linguistic entities[7]. Morover it shares several features with the dispositional theory of belief set forth by Ramsey. First, beliefs are, as Ramsey said, “maps by which we steer”: the total belief state of an agent is such a map[8]. It is the functional role of a belief, together with its location in a set of possible worlds which determines its content. Second the account can be supplemented in order to incorporate the fact that beliefs can have degrees of confidence: an agent can be inclined to take himself to be located in one subset of possible worlds rather than another. Third, it is compatible with Ramsey’s pragmatist theory of belief-content: the content of a belief that P is given by the actions for whose utilities the truth of P is a necessary and sufficient condition. In other terms the truth conditions of a belief are its utility or its success conditions. [9]  Fourth, it accommodates the phenomenon of tacit or unconscious beliefs: since the propositions which are the objects of belief are relatively indeterminate entities, they can be consciously entertained or not: the fact that they are present to the mind of their beholders does not matter. As Ramsey says, “the beliefs that we hold most strongly are often accompanied by no feeling at all”, and “my belief that the earth is round [is a belief of which] I rarely think of, which would guide my action in any case to which it was relevant”[10].

Stalnaker’s possible world version of the dispositional-functionalist view, however, faces two well known difficulties. First, as we already saw, it gives us only a very coarse-grained analysis of the content of beliefs: qualifications have to be made to take into account the problem of indexical beliefs (I do not locate myself only in the worlds in which Pascal Engel believes  there are mines in a field, but also in the worlds in which I  believe that there are mines in this field), to take into account that agents are not always rational and can sometimes switch between different beliefs worlds. Second, there is the difficulty known as the problem of deduction: if a belief content P logically entails a belief content Q, then if a subject believes P, he must also believe Q, which is by definition true in the set of worlds in which P is true. If this is so, everyone believes all necessary truths, and everyone behaves as if the necessary truths are true. But it is a commonplace that someone can, say, believe that a triangle is equilateral, without believing that this triangle has equal angles. Moreover everyone behaves as if the necessary truths are true. In other words, with necessary truths we steer by the very same maps. Here again the pragmatic functional or dispositional account individuates belief contents too coarsely[11].

At this point, we could doubt that the possible world analysis of belief content gives us the right account, and we may prefer to revert to Marcus’s analysis of the object of beliefs as Russellian propositions or states of affairs. But the problem of deduction is going to be still with us, for an agent who believes that Hesperus is Phosporus will be in relation with the same Russellian proposition or state of affairs as the proposition that Hesperus is Hesperus, and an agent who is in relation with a state of affairs which entails the existence of an other state of affairs will be in relation to both, and indeed with all the necessary states of affairs. Hence (D) does not fare better, on that score, that Stalnaker’s account.

Before being tempted to move here to conceptions which would revert to the idea that we need to construe the belief-relation as a relation to a sentence, or to an internal representation having some appropriate characteristics (such as nomological covariance, or an appropriate evolutionary history, say), or to the use of such notions as Fregean modes of presentation, we should pause a little bit to see whether a dispositionalist account like (D) or its possible world alternative is really so bad. It is not so bad, for it captures at least the fact that, given a certain desire, a belief in a necessary proposition P or another Q represent the same set of worlds. Being an equilateral triangle and having three equal angles are not two different ways things might be, and in that respect, the possible world analysis captures a legitimate notion of content.[12]  Actually this is just what Marcus’s analysis captures too: for someone who believes that Hesperus is Phosphorus and someone who believes that Hesperus is Hesperus do  believe the same thing, given that Hesperus is Phosphorus. This is what underlies her apparently surprising claim that someone cannot really believe that Hesperus is different from Phosphorus, for this is an impossible state of affairs. Of course the problematic feature of both the possible world analysis and, for that matter, of (D) is that although both beliefs, given the desires and the actions of the subject represent the same thing, they do not represent them in the same way, for a given subject, or for two different subjects, and this is just what the problem of intentionality is about. But at least the following is true: if an astronaut desires to fly to a planet and flies to Hesperus, we have a prima facie reason to ascribe to him the belief that Hesperus is a planet, even though he might represent to himself Hesperus as Phosphorus. In other terms, as Stalnaker says, the functional-dispositional view captures at least the negative fact that the agent does not locate himself, behaviourally, in the possible worlds in which it is false that Hesperus is Phosphorus. We would not ascribe to him this belief if he were not disposed to act as if it were true. In this sense, we can say that the dispositional account is the minimal notion of belief content: a  belief the P is always at least  a disposition to act as if P were true, no matter how richer the belief set of an individual might be. The first mark of believing is to behave appropriately, although there can be other more complex modes of believing. Therefore (D), or Stalnaker’s possible world analysis, are correct in this sense: belief is always a disposition to act as if a certain content were true, or as if the agent were located in the appropriate worlds, or as if the appropriate state of affairs obtained, although this does not exhaust the phenomenon of believing, since it is always to behave as if plus  a certain mode of believing.[13]

 

II

 

     What might these extra or richer features be? The natural suggestion is that they might be sentences or linguistic features, for we, humans, think most of our thoughts in words. The  familiar idea is then is that to believe that P is to be disposed to assent, either to an internal representation meaning that P, or to a public sentence “P”. Let us call this the assent theory of belief, where assent is always assent to a sentence or a sentence-like representation[14]. It is, however, important to distinguish a weaker and a stronger form of this claim. The weaker claim says that among the acts to which one might behaves as if P were true there are acts of assenting, although other forms of behaviour may justify the ascription of belief. The stronger claim is that a disposition to assent is not only a sufficient condition for belief, but a necessary one as well: there can be no belief without assent. The weaker claim can be accommodated by the dispositional-functional conception, for it says that in addition to dispositions to behave in various ways, people have dispositions to assent to sentences which express their beliefs. To have a belief is one thing, to be able to report the belief in a sentence or to assent to a representation is another thing. This distinction fares well with the distinction between explicit and implicit belief, for one can implicitly believe that P without having the means to express this belief.[15] But the stronger claim says that it is the very assent, or the disposition to assent which creates the belief. Hence it does not exist in its implicit form before it is explicitly recognised by the cognitive system of the agent (or simply written in the “belief box” without being recognised consciously)[16]. An obvious advantage of the strong assent theory of belief is that it eschews the problem of deduction: if someone believes that P, and P entails Q, it does not follow that he believes that Q, for the subject does not assent to the second sentence when he assents to the first: he needs not recognise or process the implication. But this advantage is lost if we reflect on the fact that since for the strong assent theory, assenting that P implies believing that P. And this is just the disquotational principle which Marcus criticizes when she objects to Kripke’s use of this principle in his puzzle. Pierre assents to “Londres est jolie” and to “London is not pretty”, and, given his bilingualism, he should assent to “London is pretty and London is not pretty”. But, as Marcus says (1993: 160) “The assent does not carry over into a belief”. Assent is distributed over conjunction, but belief is not. The same feature, as Marcus notes, is exemplified in the lottery paradox: given a lottery with one million tickets, one can assent to the sentences “Ticket 1 will not win”, “Ticket 2 will not win”, without believing that the conjunction “Ticket 1 & ticket 2 & …ticket one million will not win”. Together with the other arguments given by Marcus against the assent theory, such as the necessity to attribute beliefs to many non linguistic creatures and the need to distinguish beliefs from their linguistic manifestations, this suggests that only the weaker assent theory is true, but not the stronger one. Thus we can grant with Marcus the fact that a number of beliefs (and of course especially human beliefs) carry over into assents, without allowing that all beliefs do so carry over, that is without making of assent the very criterion of belief.

         But this won’t solve the problem which led us to invoke the notion of assent or the ability to express one’s beliefs in words: for the fact that beliefs are not only behavioural maps by which we steer, but can also be linguistic representations of those maps is precisely invoked in order to reduce the coarseness of grain of the map conception. The very reason why language-centered theories seem to give us a finer grain of belief contents is that taking beliefs to be relations to sentences captures better the way people take things to be. The linguistic shaping of thoughts allows us to distinguish the subjective aspects under which we apprehend thoughts which are otherwise related to the same sets of worlds, or to the same states of affairs. And such aspects, be they linguistic or not, can be behaviourally relevant (remember the difference between my belief that P.E. is attacked by a bear and my belief that I  am attacked by a bear). So, how do we locate ourselves in our belief-worlds?  Can we preserve the picture that there is nothing essentially linguistic about belief  and grant that most of our beliefs are specified through language?

We can adopt, at this point, two different strategies (although they are in many respects complementary). One, to which I have already alluded above, would be to try to specify the various ways in which we can believe one thing (be related to a certain state of affairs) under a certain mode of presentation, and another thing under another mode, and hence have different beliefs, or to specify the various manners by which our beliefs can represent things. This involves spelling out a particular theory of belief content. There are indeed may ways of doing this, among which a Fregean theory which associates each concept or expression with a particular sense or mode of presentation is only one of them. There are many other competitors on the market: informational semantics, conceptual role semantics, teleosemantics, to name a few. The other strategy, which I prefer to take, consists in taking the problem of content for granted, and to try to account for differences in beliefs not in terms of their contents, but in terms of the kind of attitudes that we have towards these contents. On this view, the fineness of grain in believing does not come only from what is believed, but also from how we believe it. Marcus was after such a distinction when she distinguished actual beliefs from mere claims to believe, when she pointed out that cases of self-deception or akrasia may manifest divided attitudes in subjects (1993: 244), and when in general she distinguished beliefs from assents. But we can look further.

 

III

 

  The dispositional-functional conception, which is much involved in our common sense ascriptions of beliefs has it that we act upon our beliefs, or that they explain our actions. But we do not simply act out of our beliefs. We also often act because we accept  certain things, take for granted certain things, or see ourselves as committed  to certain things, in addition to our believing certain things, and sometimes in spite of our believing certain things . For instance a teacher may give a good grade to a student’s paper, and thus accept that the paper is good, in order to encourage him, although he believes that the paper is not really worth this grade. Or someone may feel that he is committed to a certain thesis, although he is not very confident of its truth. A mathematician who intends to prove a proposition by reductio ad absurdum may take for granted that the negation of this proposition is true, in order to show that it is contradictory. Such psychological states as accepting that P, taking it for granted, premising it, committing oneself to it, or postulating it, do not seem to be beliefs proper, although, we could say, they arise in the neighbourhood of beliefs and are belief-like attitudes. A number of writers, and in particular Stalnaker, Lehrer, Cohen, and Bratman[17], have emphasised the distinction between beliefs and acceptances, mainly on the following points. (1) Whereas beliefs are involuntary, acceptances seem to be voluntary, or at least seem to occur in the context of practical, rather than theoretical reasoning: acts of acceptance are performed with an intention to achieve a certain result, rather than to have as outcomes the truth of certain propositions. (2) Acceptance-states, for this reason, are contextual, and relative to the aim at stake, whereas beliefs are not.(3) Acceptance is qualitative or categorical, whereas belief is graded and subject to degrees. Two further features seem to derive from these points. Whereas beliefs can be unconscious of tacit, acceptances seem to be explicit and reflective. A proposition which I accept seems to be in need of being present “before the mind”. And states of acceptance, for this very reason, seem to be relations to linguistic entities, such as sentences. How, for example, can I accept, for the sake of a reductio ad absurdum proof, that P is not true, without considering explicitly the sentence “P” and what it means? Cohen, in particular, takes acceptance to be dependent upon language. For these reasons, it is very tempting to assimilate acceptances to assents, which we have defined as relations to sentences[18]. But this line of thought creates a dilemma: either acceptances as assents are mere manifestations of belief, in which case it is not necessary to distinguish them from beliefs proper, and one does dot see why the alleged differences (1)-(3) mark a real distinction, apart from the fact that acceptances are beliefs fleshed out in linguistic form, or acceptances as assents are states which are different from beliefs, which implies automatically the falsity of the stronger assent theory of belief (if assent is different from belief, the former cannot be a necessary and sufficient condition of the former) and the dubiousness of the weaker thesis, since it precisely takes assent to be an explicitation of beliefs which can otherwise be implicit.

        I want to reject the dilemma: acceptances are identifiable neither to beliefs nor to assents, although they share a number of properties with both, since they enjoy a status which is intermediary between the two. Let us start with the differences and similarities between belief and acceptances, before envisaging the connections and differences between acceptances and assent. I take up the points (1)-(3) above in turn.

(1) Accepting that P, unlike believing that P, is an intentional act, something like to take the decision to hold a proposition. This constrast strongly with the essential unvoluntariness of belief[19]. In particular, writers on acceptance frequently claim that one can accept a proposition without believing that it is true, and even while believing that it is false, hence contrary to the evidence for its truth. If we take belief to be essentially a passive state shaped by evidence, the possibility of such acts of acceptance marks a sharp difference from belief. Conversely, if we take acceptances to be cases of deciding to believe or of successful will to believe, then the existence of such psychological states would be an argument for identifying  acceptance and belief, when belief is, on this view, voluntary. I shall not consider this converse line of argument, for I do not think that there genuine cases of voluntary belief, although I shall not deal with this matter here[20]. Now if we focus on the standard examples of acceptance without belief, such as Cohen’s case of the lawyer who accepts, for professional reasons, that his client his guilty although he believes that is guilty, it is not obvious that the lawyer decides to believe, or brings it about that he believes that his client is innocent. On the contrary, he maintains his acceptance as a practical postulate, or as a provisional premise, for the sake of the trial, without withdrawing his belief to the contrary. Most of these cases involve what we may call pretendings to believe, or simulated beliefs, in general in social contexts, and most of the time they involve some sort of conflict between epistemic and practical norms of belief formation. But such cases of accepting that P in spite of the evidence that justifies the belief that P, although they dramatically reveal the difference between the two kinds of states, are not necessarily central cases of acceptances. Many cases of acceptance are cases where the agent still has evidence for what he accepts, although the evidence is in some respects insufficient. Take Bratman’s (1992) example of someone who projects to build a house and estimates the cost at a certain price, but who nevertheless accepts that the price will be higher than his estimate, for he wants to contract a loan, and does not want, in case the actual price would be higher, to have an a lower funding. His acceptance is distinct from his belief (if he had to bet on the price, he would not evaluate the price that high); still he accepts the higher price on the basis of his evidence[21]. Or take the cases of acceptances as premising in the course of a theoretical reasoning: they differ from beliefs, for they involve taking for granted a certain proposition P which one may otherwise only believe to a certain degree, but they are, like beliefs, shaped by a concern of evidence, and they “aim at truth” just as beliefs do. In other terms, if justification and truth are norms of belief formation, as the “evidentialist” thesis has it against voluntarism[22], acceptances need not diverge from these norms. What they imply, however, is that the subject has a more conscious grasp and recognition of these norms than in the case of belief.

This remark gives us a clue for understanding feature (2) above, the context sensitivity of acceptances. A subject who accepts a proposition in the course of a reasoning does not necessarily withdraw, for the purpose of his reasoning, his epistemic end of searching after truth. But this end becomes for him more conscious than when he simply forms the corresponding belief and acts upon it. In this sense, there are two ways to reason, for instance, according to a modus ponens rule of inference: one consists simply in drawing, from the belief that P and the belief that if P the Q, the conclusion Q, by acting as if it were true (for instance by refraining to do things which would imply that Q is not the case); the other consists in acknowledging the rule itself, and take it as a reason which validates the conclusion. The outcome, in both cases, is the same – it is the entertaining of the belief that Q – but it is not reached in the same way. Thus psychologists of reasoning distinguish between the psychological process of deduction which individuals perform more or less irreflectively and the process of meta-deduction which “prepare the way for the development of self conscious methods for checking validity”[23]. I suggest that acceptance, most of the time, belongs to this (mostly, see below for a qualification) reflexive stage of inference.

(3) The last contrast, between belief as an attitude susceptible of degree, and ideally measurable through subjective probabilities, and acceptance as a qualitative attitude to propositions taken to be true or not, seems to be the stronger one. Stalnaker (1984,ch.5) introduces precisely this distinction in order to deal with the lottery paradox: I can certain believe to degree 0,99999, say, that respectively ticket 1, ticket 2, …, ticket 10000, will not win, but  my beliefs need not carry over to acceptance that no ticket will win (1984: 91). Acceptances, being, unlike beliefs, categorical states, are governed by the principle that to accept that P and to accept that Q entail that I accept the conjunction P&Q. Similarly, although I should accept that Q if I accept that P, and P entails Q, it does not follow that the corresponding entailment holds for my beliefs (although it holds for what I should believe). The distinction is sufficient to block such contradictions as those that are exemplified in the lottery or in the logical omniscience examples. But that does not show that there has to be a huge gap between belief an acceptance. It does not follow, if we suppose, like Bayesian decision theorists, that our beliefs have subjective degrees of probability, and that we act on them together with a principle of maximising utility, that our acceptances should not mirror our beliefs. Just as in deductive reasoning accepting a proposition involves becoming reflectively conscious of it, in probabilistic reasoning accepting a proposition involves using consciously the Bayesian standards that we use tacitly[24]. This may involve, as in the lottery case, some conflicts  between our usual deductive standards and the Bayesian ones, but such conflicts are to be expected at the level of acceptance.

From what precedes, then, I conclude that although acceptances are distinct from dispositional beliefs, they need not be states which exclude beliefs and which are totally unrelated to beliefs. Even in the dramatic cases were they diverge from belief, they arise in the conscious evaluation of diverging norms for belief: deductive vs inductive, theoretical vs practical. But now we face a further difficulty. For is seems that the characterisation that I have given of acceptance, as belief made conscious, makes it close to assent. For how could we accept propositions in a reflective manner without considering their linguistic clothing in sentences. Certainly, the description that I have given above of reasoning, in the acceptance mode, from a modus ponens style of inference, seems to involve considering sentences. Moreover acceptance as I have characterised it, seems to be made possible by the capacity to reach second-order beliefs, beliefs about beliefs, which are often said to be made possible only through language.  It is an issue, in the theory of mind literature as well as in ethology, whether meta-propositional thought is made possible by language, but this thesis seems to be immensely plausible.[25] It is, therefore, tempting to come back to the assimilation of acceptance states to states of assent, and given that I have emphasised the similarities between belief and acceptance, it runs the risk of bringing back a language-centered view of belief. But I think that we can grant that what we might call the acceptance system in the cognitive economy of an individual is related, both in its formation and in its activation, to the capacity to process and to consider sentences, without granting that acceptance is essentially a linguistic state, mainly for two reasons. First, if acceptance is mostly a conscious state, it does not follow that all conscious reasoning through acceptance goes by explicitly considering sentences. If I form a plan, for instance, to come back home by passing in front of a library and if I accept, in the course of my plan, that I shall pick up a book if the library is open, my conscious thoughts need not involve the consideration of sentences. Similarly for many cases of practical reasoning. Second, although I have tended above to assimilate acceptance to a conscious state, it need not be so: as Cohen says, there might me tacit premising, and tacit acceptances as well[26]. This seems to be acknowledged by psychologists of reasoning. For instance when Johnson-Laird and Byrne (1991) claim that propositional reasoning goes by the formation of mental models and by the use of a semantic principle that an inference is valid if it does not have counter-models, they can follow tacitly this rule and the procedures of mental model formation, and hence accept them, without consciously doing so. Nothing seems to prevent us from saying that there is, in such cases, a tacit knowledge of normative principles[27]. But what, then, would be the difference between these tacit acceptances and the tacit beliefs that we have granted on the dispositional view? Surely such tacit acceptances have in common with the tacit beliefs that they are dispositions, and that we act upon them as if they were true. But what they do not have in common is that they are dispositions to reason, and not simply to act, upon a certain pattern of beliefs, such as, for instance, the modus ponens pattern. Hence their importance in inferences. What marks their difference with mere dispositional beliefs is that they are dispositions to reason, not simply to act.[28] Thus the active features of acceptances do not imply the falsity of an overall functionalist picture of the mind: they are functional states. Only a narrow kind of functionalism, which would insist on the essential passivity of belief, would contradict the acknowledgement of such states.[29]

If this is correct, then, acceptances share both with beliefs certain dispositional features, which relate them to acting as if certain propositions were true, and with assents certain other features, which relate them with the explicit consideration of linguistic propositions. But what I have tried to suggest is that they reduce to neither of them, and enjoy a intermediary status. Assents and acceptances are cases of dispositional beliefs plus something, where the “something” is different in each case. Instead of a simple difference between belief and assent, as that upon which Ruth Marcus lays correct emphasis, we should have a complex series of distinctions, between beliefs and acceptances, between acceptances and assents, and between beliefs and assents. But the distinctions do not mark, in many cases, sharp oppositions. All these states lay on a continuum. In many respects it seems to be just a matter of terminology, for sometimes it may be more economical to say that we are simply dealing with “beliefs”. But in other cases failure to sort out the various kinds of states leads to a distorsion of the facts. This is why doxastic life is so complex.*


REFERENCES

 

 

Braddon-Mitchell, D & Jackson, F. 1996 Philosophy of Mind and Cognition, Blackwell, Oxford

Bratman, M, 1992 “Practical reasoning and acceptance in a context”, Mind, 102, 1-15

Carruthers, P., & Smith, P.K., (dirs.),  (1996), Theories of Theories of Mind,  

                     Cambridge, Cambridge University Press.

Cohen, L.J. 1992  An Essay on Belief and Acceptance, Oxford, Oxford University Press

Dennett, D. 1978 “How to change your mind”, in Brainstorms, Bradford Books,

       Cambridge, Mass

de Sousa, R. 1971 “How to give a Piece of your Mind, or: the Logic of Belief and

                 Assent”, The Review of Metaphysics, 35, 52-79

Engel, P. 1995 “Les croyances”, in D. Kambouchner, ed. Les notions philosophiques,

                       vol. II Paris, Gallimard, 9-101

      1996  “Le rôle de la croyance dans l’explication de l’action”, in Petit, J.L.

                  ed. Les  neurosciences et la philosophie de l’action, Paris Vrin, 327-340

      1997 a “ Dispositions à agir et volonté de croire”, in Proust, J. & Grivois, H.

                  eds, Subjectivité et conscience d’agir, Paris, PUF , 115-137

      1997 b “De modalibus non disputat asinus”, critical review of Marcus 1993,

                 Dialogue, XXXVI: 157-69

      1998a  “Believing, Accepting, and holding- true”, Philosophical Explorations,

                II, 1, 140-151

      1998 b “Sentences, Statements, and propositions” , in E. J. Craig, ed. 

                   The Routledge  Encyclopedia of Philosophy, Routledge , London

      to appear ed. Believing and Accepting, Kluwer, Dordrecht

Frankish, K. 1998 a “A matter of opinion”, Philosophical Psychologyi, 11, 4, 423-442

                   1998 b “Virtual Belief”, in Carruthers,P. & Boucher, J. eds, Language and

                              Thought, Cambridge, Cambridge University Press, 249-269

Johnson-Laird, P. & Byrne, R. 1991 Deduction, L. Erlbaum, Hove and London

Kripke, S. 1979 “ A puzzle about Belief”, in Margalit, A. ed. Meaning and Use,

Dordrech, Reidel, 234-283

Lehrer, K. 1990  Knowledge, Westview, Boulder, Colorado

                to appear “Belief and Acceptance revisited” in Engel ed. to appear

Marcus, R. B, 1981 “ A proposed solution to a puzzle about Belief “ , Midwest

                            Studies in Philosophy, VI. 501-510

    1985 “Rationality and Believing the Impossible”, The Journal of Philosophy,

                          LXXX, 6, 321-338

1986  “Some Revisionary Proposals about Belief and Believing”, Philosophy  and Phenomenological Research, L, 133-153

    1993 Modalities, Oxford, Oxford University Press

    1995 “The Anti-naturalism of some Language-centered Accounts of

                         Belief”, Dialectica, 49, 2-4, 113-129

Peacocke, C. 1992 A study of Concepts, Cambridge Mass., MIT Press

Pettit, P. 1998  “Practical Belief and Philosophical Theory”, The Australasian

                   Journal of Philosophy , 76, 1, 15-33

Ramsey , F. (1927)  “Truth and Probability”, Philosophical Papers, Mellor ed.

      Cambridge University Press, 1991

Salmon, N. & Soames, S. eds 1988 Propositions and attitudes, Oxford, Oxford University Press

Schiffer, S. Remnants of Meaning, Cambridge, Mass, MIT Press

Stalnaker, R. 1984 Inquiry , Cambridge Mass, MIT Press

Williams, B 1971  “Deciding to Believe” in Problems of the Self, Cambridge,

           Cambridge University Press

Williamson, T. (1996) “Knowing and Asserting”, The Philosophical Review, 105, 489-523

Whyte, J. 1991 “Success Semantics”, Analysis, 50, 159-157



[1] Marcus 1981;  1983, 1990 (both of these quoted from her 1993); 1995.  These essays, written in her caracteristically vigorous and lucid style, may well be as pioneering in their field as her more well-known work on modality. For instance, her account implies that there is much more symmetry between knowledge and belief than it is usually said, an insight which has recently be rediscovered by Williamson (1996); see for instance 1993: 144.

Ruth Marcus’s uncompromising sense of what good philosophy should be and her unfailing search after truth have been for me a model. I dedicate the present paper to her with admiration and affection.

[2] Kripke 1979.

[3] In my critical review of Marcus 1993 (Engel 1997b) I emphasised this point. I try here to develop it. See also Engel (1995), Engel (1996  and 1997a) , Engel  to appear

[4] Marcus 1995: 126; 1993: 241

[5] Brathwaite 1932-33, Ramsey 1926

[6] for such accounts, see for instance the papers in Salmon and Soames 1988. for the difficulties, see in particular Schiffer 1986 and for a summary of the difficulties Engel 1998 b

[7] Marcus 1993 (1986) : 147, note 5, notes this point, in the only passage that I know of where she refers to Stalnaker’s theory.

[8] For an explicit connexion between the map image and the possible world analysis, see Braddon-Mitchell and Jackson 1996: 179-190

[9]  For a defense of the Ramseyian pragmatist theory of belief content, see Whyte 1990 ( who calls it a “success semantics”)

[10] Ramsey 1926, p.65 ; p.68 respectively.

[11] Se Stalnaker 1984: 82

[12] Braddon-Mitchell & Jackson 1996: 192.

[13]  In this sense, Pettit (1998: 17) says : “The first mark of believing something- if you like the criterion of belief- is that you are disposed to act appropriately. Behavioural-plus modes of belief satisfy this mark and will be distinguished from mere behavioural belief by the fact of involving something else as well.” And he adds, just along the same lines as Marcus: “ This is not an uncontentious line to take, of course. It means that all we have to say, for example, that someone who claims to believe something that is necessarily false – someone who assents to a necessarily false assertion – cannot really believe it, they can only believe that the words accepted express a truth.”( ibid. note 1).

[14] Pettit 1998 uses a different term, and calls this the “judgmental” conception of belief. It should be clear that assent, in the sense in which I am taking it, is not simply assent to a sentence, for otherwise parroting, or mere the “holding-true” of a sentence which one does not understand would be belief.  The assented sentence has to be understood, and thus must be have a meaning for the subject. It has to be what the sentence expresses, either in the language of the subject, or in some language into which it can be translated, hence it must be something like a proposition.

[15] Braddon-Mitchell and Jackson 1996: 192. Marcus makes remarks to the same effect in 1993: 242-243

[16]  As Braddon-Mitchell and Jackson (ibid.) aptly remark, this  does not imply that views according to which belief is assent to a internal sentence are unable to acknowledge the existence of implicit beliefs, but it implies a contrast between an internal sentence theory of implicit belief and a dispositional one: for the latter all  beliefs are potentially implicit and are already there in the cognitive system, whereas for the former, implicit beliefs are those which can be derived, more or less easily, from a core of internal sentences. In other terms, for the dispositionalist you do believe that elephants don’t wear pyjamas, whereas for the internal sentence theorist you don’t really believe it until someone attracts your attention to it.

[17]  Most notably Stalnaker 1984, ch. 5, Lehrer (1990), Cohen (1992), Bratman (1992). For a review of this literature, see Engel (1998) and Frankish’s two remarkable papers (1998a, 1998b).

[18] And indeed many writers seem to do so. For instance De Sousa (1971) in a pioneering paper on this topics, seems to call “assent” what we call here “acceptance”, and Dennett (1978) talks of “opinions” as a species of  linguistic assentings.

[19] Locus classicus: Williams 1971.

[20] For essentially the same reasons as those given by Williams (1971). The fact that acceptances involve a voluntary element does not imply, in my view, that voluntarism about belief is true. The distinction between acceptance and belief implies only a weak form of voluntarism, although I shall to try to ague for this here. For more on this issue, see Engel 1997 a.

[21] Similarly, Kant’s ( KRV, A 824/ B 852) example of a physician who accepts that his patient has tuberculosis, although he is unsure, on the basis of the symptoms that it is actually tuberculosis: he prefers to err on the side of prudence, but his acceptance is also justified by his evidence

[22] Evidentialism is usually defined as the thesis that beliefs are involuntary and only subject to epistemic norms, whereas voluntarism combines both the psychological thesis that there can be intentional acts of believing and the normative thesis that epistemic norms can be, sometimes, preferred to practical or prudential norms.

[23] Johnson Laird & Byrne 1991: 147

[24] The point is well made by Frankish, 1998: 433.

[25] See for instance, among others in a vast literature, Carruthers & Smith 1996

[26] Frankish 1998 b, develops a whole theory of acceptance, as emulation or simulations of first-order dispositional beliefs, which he calls virtual beliefs, of which tacit acceptances are only a part. I take the remarks given here to be very much in line with his, although I do not deal with the notion of simulation.

[27] This marks a difference with usual accounts of tacit knowledge, which ascribe such knowledge about a descriptive theory (of grammar, say). what Peacocke (1992) calls possession conditions for concepts (especially logical concepts) are both normative in this sense, and not necessarily explicit.

[28] What I call implicit acceptance in this sense are quite close to what Pettit (1998) calls “practical beliefs”, which he defines precisely as “dispositions to reason according to a certain pattern.”

[29] It is characteristic of Stalnaker’s account of acceptances (1984: ch.5) that he defines them as functional states alongside with beliefs. The same point is stressed by Lehrer (forthcoming).

* Many thanks to Keith Frankish for his remarks on this paper