Dretske and the causality of reasons                         back


In his work on reasons Dretske argues that reasons are only worthwhile for having them if they are causally relevant for explaining behaviour, which he elaborates in his representational theory of explanation. The author argues against this view by showing that there are reasons that are relevant for explaining behaviour but not causally relevant. He gives a linguistic foundation of his argumentation and shows that Dretske’s representational theory cannot explain human actions because man does not only perceive things that have already meaning but also assigns meanings to what (s)he perceives and that therefore reasons are fundamentally different from causes.


Dretske and the causality of reasons

Henk bij de Weg

How can it be that a thing has a meaning and that the fact that this thing has this meaning can explain certain effects or is at least relevant for the explanation of these effects? This is Dretske’s main question, when he asks himself what reasons are. When he asked this question, Dretske had two things in mind. First, the meaning as such must be relevant for the explanation. "A soprano’s upper-register supplications may shatter glass, but their meaning is irrelevant to their having this effect" (Dretske 1988, 79). Even if the sounds had no meaning, the effect would be the same. However, there are cases where the meaning of a thing is explanatorily relevant, and it is these cases that Dretske’s theory of the causal role of meaning refers to (id., 77-80). Second, explanatory relevance must be for Dretske causal relevance. In "Reasons and Causes" he puts it like this:

If beliefs and desires are not causally relevant to behavior, I, for one, fail to see why it would be worth having them. We need beliefs and desires because our wanting this and believing that, besides being our reasons for doing what we do, are – sometimes at least – the reasons why we do it. If reasons aren’t causes, one of the chief – indeed (for certain people) the only – motive for including them in one’s inventory of the mind, vanishes. They are no longer capable of doing the job – actually getting us to do what they justify us in doing – that was their primary excuse for existing (Dretske 1989, 1; italics by Dretske).

And a few pages further down Dretske says it even stronger:

Any theory of meaning that doesn’t make a thing’s having meaning into a causally relevant property of the thing (and hence the fact that it has meaning into an explanatorily important fact about the thing) is a theory of meaning that can be rejected at the outset (id., 5).

Meanings that determine behaviour are for Dretske reasons in the first place. Especially, they are beliefs and desires. In this article, I want to discuss Dretske’s view on the causal role of reasons. In § 1, I ask whether reasons as meanings really need to be causally relevant factors in order to be valuable for explaining behaviour. How, according to Dretske, reasons are causally relevant for explaining behaviour, and why I criticize this view is set forth in § 2. In 2.1, I show that Dretske’s concept of cause (but not only his concept of cause) has in fact two different meanings, which in Dutch are called "oorzaak" and "reden". In 2.2, I formulate my criticism against Dretske’s representational model. This model is meant to show the role of meaning for the explanation of behaviour, and maybe it is appropriate for showing the role of meaning assigned to physical phenomena. However, social phenomena are substantially different, for they are pre-interpreted phenomena, and this fact causes insurmountable explanatory problems for Dretske’s model. In 2.3, I apply my analyses to Dretske’s distinction between triggering and structuring causes. By this, I hope to make clear that it is not sensible to reject a theory of meaning "at the outset", if the meanings enclosed are not causally relevant, for they may be explanatorily relevant in a different way.

Before starting my analysis, I want to say how I use the word "explanation". For many authors explanation is synonymous with causal explanation. For me explanation is broader and it includes also understanding ("Verstehen") and detecting functions. I shall make it explicit, when I mean causal explanation, unless it is clear from the context.

1. The causal relevance of reasons

The quotations in the introduction show that causal relevance of reasons implies for Dretske two things: 1) If reasons are not causally relevant for behaviour, they are not relevant in a different way either. For if reasons are not causally relevant, although they are relevant in a different way, then we might suppose that Dretske would at least attach some value to having beliefs and desires (i.e. reasons) in that case, and he would not reject a theory that ignores or rejects their causal relevance at the outset. 2) If we talk about the reasons why we do something, this "why" has a causal meaning.

Before entering into the question whether these two postulates are tenable, I want to elucidate Dretske’s concept of reason. For Dretske, reasons are "those content-possessing mental states (belief, desire, fear, regret) we invoke to explain one another’s behavior" (Dretske 1988, 79). Particularly, the agent’s reasons are the cognitive factors and conative conditions that steer his behaviour. The cognitive factors C are generally called by Dretske "beliefs", and their function is ".… to indicate the presence of those conditions that, if the right motivational state is present, will lead, other things being equal, to M" (id., 105) with M being what is done by the agent, i.e. his behaviour or action. However, having a belief is not sufficient for M taking place. In addition, there must be a conative condition or what Dretske generally calls a "desire", i.e. a certain motivational state (D). Basically, the cognitive factors and the conative conditions determine together the agent’s behaviour, and so they are the reasons for this behaviour (id., 105-107). By way of illustration, Dretske gives the example of a thermostat. An indicator registers the room temperature (cognitive factor), but it turns the furnace on only if the room temperature is below the desired temperature (conative condition). Or we have learned a rat to press a bar in order to get food, but later, when seeing the bar (cognitive factor), the rat will not press it, unless he is hungry (conative condition).

Against this background, I want to analyse an example for testing Dretske’s two postulates. Take the case of a friend of mine who has called me up asking whether I can come to help him. I promise to come as soon as possible, I take my coat, walk to the shed, and take my bike. Just as I want to ride off, my wife calls me and asks me to post a letter.

What I do now can be described in two ways: 1) posting my wife’s letter; 2) going to my friend. Let us look first at 1). If we apply Dretske’s theory, I think the cognitive factor here is my belief that my wife wants me to bring the letter to the postbox (because of her question). I want to do my wife a favour, and so I have a desire (conative condition) for really posting the letter. This analysis seems to explain my action "posting the letter" (M). However, it is not the whole story. This becomes clear if we look at my "second" action: going to my friend. This action can be "explained" in the same way as the "first" one, but that is not what matters here. I want to examine the relation between both actions. The fact is, if I had not gone to my friend, my wife would not have asked me that question, and I would not have posted the letter. Maybe she had posted it later herself, when she went out. So my going to my friend is indispensable for my going to post the letter: I bring the letter to the postbox and I do it because I go to my friend. My going to my friend is therefore a relevant "explanatory" factor of my action "posting my wife’s letter": it is a consideration that I can advance, if someone asks me why I have posted the letter. "I was about to go to my friend and then my wife asked me to post this letter and so I did" is what I can explain to the judge, when she asks me in court what I did on the scene of the crime. Accordingly, this consideration is a reason as described by Dretske, namely a "belief". But is it also a causal relevant "explanatory", namely a cognitive, factor for my action "posting the letter"? Dretske says with right that cognitive factors can be causally effective only if there is an accompanying conative condition, or "the right motivational state", as Dretske formulates it in the quotation above. As he says it elsewhere: "a desire for whatever reward or reinforcement promoted C into a cause of M" (id., 105). As just said, the conative condition (desire) in my example is that I want to do my wife a favour. However, the consideration that I am to go to my friend does neither refer to a circumstance that can fulfil my wanting to do my wife a favour (just as seeing the bar can do in the case of the hungry rat), nor is it a cognitive factor that is or can be fulfilled by this conative condition (becoming hungry if one smells food). As Dretske puts it, it is not an "internal indication of the appropriate stimulus conditions" (id., 113n). In order to fulfil this conative condition, we need another cognitive factor that does indicate the appropriate circumstances, in this case that my wife calls me and asks me to post the letter. I go to my friend because he called me up and because I want to do him a favour. It is not my going to my friend but my wife’s question that is the causally effective reason for my action of posting the letter. At least it is in the sense of "reason" given by Dretske. However, in the presence of another cognitive factor, my going to my friend becomes a relevant reason for doing my wife a favour, i.e. to post a letter for her, and this is what happens in my example. So, what we have here is the case of a (cognitive) factor that is a relevant reason for an action but not a causally relevant reason in Dretske’s sense. As it happens, reasons can also be used for explaining actions to other people (and to oneself!), i.e. for making something understandable, or for justifying.

In conclusion, we can say that reasons can be relevant for the "explanation" of what I do, i.e. my actions, without being causally relevant for it. In spite of this, reasons give an answer to the question why I do that.

2. Dretske’s theory of representation

Despite of the conclusion of §1, it is still possible that there are causally relevant reasons. In this section, I want to challenge also this point in a critique on Dretske’s theory of rationality. I start from the following quotation:

(1) To say that the thermostat turned the furnace on because the room temperature dropped to 68o is like saying that the duke stood up because the queen entered the room. (2) To say that an object, whether it be a person like the duke or an instrument like a thermostat, did something because of some (cotemporaneous) fact about its external surroundings is to give the triggering cause of the behavior. (3) In the duke’s case, the queen’s entry into the room caused in him a belief that the queen was entering the room – an internal state that, if things are working right, represents the queen’s entry into the room. In the thermostat’s case, the fall in room temperature brings about an internal condition – a condition that, if things are working right, represents the temperature in the room. In both cases this representation is harnessed to a motor control system. In the duke’s case it brings him to his feet. In the instrument’s case it switches on the furnace. (4) And in both cases we can ask about the structuring cause of this behavior – why the internal representation has this effect rather than some other effect; why this process rather than some other process is triggered by the external stimulus. When we seek an explanation of behavior in terms of the agent’s reasons, we are, I submit, always looking for a structuring cause (id., 50; italics by Dretske; my numeration).

I have quoted this passage at length, for here we find all elements of Dretske’s causal theory of reasons. The central point in it is Dretske’s analogy between, what he calls, the "behaviour" of a thermostat and the "behaviour" of a certain person, in this case a duke. I would rather say that there is, at least according to Dretske, an analogy between the movements of a mechanism that reacts automatically on outside changes and the actions of a person in relation to his environment. Since Dretske equates them, he runs into difficulties, as we shall see, and that is the main point of my argument against him. Before formulating my critique in detail, I want to sum up four points of agreement between both kinds of "behaviour" in this quotation (cf. the numbers there):

1) The "because" that shows why a thermostat reacts on a temperature change agrees with the "because" why the duke stood up for the queen.

2) Both becauses refer to a triggering cause.

3) In both examples, an internal condition is brought about that represents the external world and that starts a process (or that is at least partly responsible for it).

4) Both what the thermostat does and what the duke does have structuring causes: the "reasons" of the "behaviour".

In my discussion of these points, the thermostat and the duke are symbols of two worlds: the world of things and the world of persons. I think that this is in agreement with Dretske’s intentions, and with the fact that he has deliberately chosen examples from both worlds, and that he wants to show how they are similar. In this way, we are back in the old discussion whether the world of relationships between physical objects is or is not to be distinguished from the world of relationships between reasonable subjects (Beckermann 1977, 19). I want to skip this problem of dualism (cf. bij de Weg 1996). Here, I want to look only at Dretske’s argumentation. Then my question is: do the agreements as supposed by Dretske really exist? In 2.1 and 2.2, I shall successively discuss 1) and 3); I will treat 2) en 4) in 2.3.

2.1. According to Dretske, there is an agreement between the "because" that indicates why a self-regulating mechanism functions and the "because" why someone acts (see the first sentence of the quotation  above). The phrase "is like saying" puts mildly what later is beyond all doubt in Dretske’s argument: that there is no fundamental difference between the first "because" and the "cause" behind it, and the second one. However, is there really no difference between the first because and the second because, or, as I will call them from no on, "because I’ and "because II" respectively (generally, so apart from these particular instances)? Here in 2.1, where I discuss the first analogy, I shall look at the linguistic aspects of the problem. In 2.3, where I discuss the third analogy, I shall link my analysis to Dretske’s causal theory.

From a linguistic point of view, the issue at stake is whether we mean the same thing with "because I" and the "because II" in the sentence quoted. I suppose that there are two concepts of "cause" behind both becauses, and the question is whether both concepts correspond. In my analysis, I make use of Gean 1966. Like Dretske, Gean supposes that reason explanations are causal explanations, but in Gean’s article the relevant linguistic aspects are more prominent than in Dretske 1988. I shall show that my criticism on Gean applies to Dretske as well.

Gean advances three arguments why reasons are causal according to him. Only the first one is relevant here. It is in agreement with Dretske’s argument in the first sentence of the quotation. It runs as follows: "… causal idioms which are appropriately used to speak of a causal relationships elsewhere are often used to request, mention or allude to an agent’s reason for action" (Gean 1966, 674). For illustrating his argument, Gean presents four sample sentences. I shall quote two of them (ibid.):1

(1) What caused him to leave the party so early?

(2) The nice weather is one of the things that caused me to come here.

What does Gean mean by these two examples? Or rather, what kind of answer is to be expected to (1)? And what kind of influence does the weather have on my coming here? The problem how to answer these questions becomes apparent if we translate (1) and (2) into Dutch. The verbal translation of "to cause" is "veroorzaken", but to talk in terms of the "oorzaak" ("cause") that the agent left the party sounds odd in Dutch, to say the least. In Dutch, one would expect that an expression in terms of a "reden" ("reason") would be used. But this shows that different things are meant according to whether one translates the sentences in terms of the agent’s "oorzaak" or his "reden". If this would not be so, "reden" and "oorzaak" would be interchangeable. That both words really mean different things becomes clear, if we answer both questions of mine. At least in Dutch, in (1) we are thinking of a physical event or an event that could not be hold off by the agent in any way: that the agent left the party was something that happened to him. Talking about an "oorzaak" there sounds so odd that it is even difficult to think of a real situation in which not also the party has come to an end (but a party that has come to an end cannot be "left"). (1) makes one think of a fire or the breaking down of the party room. But maybe we have to think of the agent’s physical collapse, so that other people have to bring him to the hospital. Or the agent was removed from the party, since he was quarrelling.2 However, we get another situation, if we formulate the question in terms of the agent’s "reden". Then the sentence sounds natural, and we expect an answer in terms of agent’s considerations plus a choice what to do: the agent could have stayed at the party (although one might wonder whether this would have been "reasonable"), but the agent made the choice to leave the party. He did not like the atmosphere there or he had been called up that a serious accident had happened to his wife.

The answer to my second question runs on the same lines. If we should translate (2) in terms of the agent’s "oorzaak", we should expect that something happened to me because of the weather, which made that I had to avoid it; here, too, it is difficult to find a "reasonable" example in which the agent does not have a choice to withstand the urge to go away. But in order to clarify the difference between "oorzaak" and "reden", in case the agent has an "oorzaak" in her mind that made her avoid the weather, she may say that she suffers from hay fever or that she cannot bear the warm weather and that this made her go somewhere, where the circumstances were better. If she refers to a "reden", the agent is maybe on a beach, when she says it, or she is making a walk in the forest, things she likes to do if the weather is warm, but which she does not need to do.

This difference between "oorzaak" and "reden" does not occur only in Dutch. In German, which is strongly related to Dutch, but also in Japanese, which is neither cognate to English nor to Dutch, there is a strict distinction between words indicating "reden" and words indicating "oorzaak".

The upshot of my discussion is this. According to Gean, it is characteristic for the nature of reasons in general that the agent’s reasons are expressed by a causal terminology, and he inferred that an explanation from reasons is nothing but a kind of causal explanation. Gean based his conclusion on the way that the word "cause" is used in English. However, this conclusion is not tenable, if one involves other languages in the analysis. Only because in English "causal" idioms are used for referring to reasons in order to formulate reason explanations, we cannot conclude that a reason explanation is a kind of causal explanation. In other languages, there is a clear distinction between "cause" and "reason". This indicates that there is a real distinction between both, or at least that a distinction may exist. And it is certainly not correct to say that causal idioms are often used to refer to reasons, as Gean does. We do not refer to reasons only in English, and there are other languages that do not use causal idioms for referring to reasons and that even cannot use them in that way. By comparing the English "cause" with its Dutch, German or Japanese equivalents, I reach the conclusion that the English "cause" is more comprehensive than these equivalents, and that it has two different meanings. First, "cause" has a meaning that is more or less similar to the Dutch "oorzaak". I want to call it the "narrow meaning" of cause or "cause in a narrow sense" (cause i.n.s.). It is the opposite of the Dutch "reden" but also of the English "reason". For in English there definitely exists a word expressing the idea behind the Dutch word "reden". However, "reason" ("reden") is not subordinate to "cause i.n.s.", but it is a concept of its own. It has a place on the same level and next to the concept of "cause i.n.s.". A reason is not a kind of cause but it expresses something different.

In addition, the English "cause" has a broader meaning. It is also used as a covering term for "cause i.n.s." and "reason". I want to call it "cause in a broad sense" (cause i.b.s.). Gean’s mistake is that he confuses both concepts. When Gean talks about "causal idioms" (cf. the quotation), he apparently wants to refer to what I have called "cause i.b.s.", but when he talks about "causal relationships" nothing else than "cause i.n.s." can be meant. But if that is true, then it is neither correct to say, as Gean does, that "causal idioms" are appropriate "to speak of causal relationships", nor is it so that use of "causal idioms" is an indication of the causal nature of reasons.

I return to the agreement that Dretske sees between, what I have called, "because I" and "because II". According to Dretske, there is no distinction between them, which he tries to substantiate with his representational theory of explanation. I think, however, that Dretske has fallen into a linguistic trap because of the external resemblance between both concepts, which made him overlook their intrinsic differences. Of course, I do not want to say that a native speaker of English cannot see these differences, but they are easily ignored, for in each case the same word is used in English. For clarifying my point of view, I want to examine the first sentence of the quotation of § 2 with the help of the analysis just made. I have split up the sentence for reasons of clearness:

A - The thermostat turned the furnace on because the room temperature dropped to 68o.

B - The duke stood up because the queen entered the room.

Instead of the omitted words "is like", I ask: is the relationship expressed by "because" in A equal to the one expressed by "because" in B? A simple way to answer this question is to replace "because" by its Dutch equivalent and then see what happens. The most obvious choice is "omdat", but in Dutch there are also two other "causal" conjunctions, namely "doordat" and "aangezien" (ANS, 655-656). "Doordat" cannot be used in sentences that clearly indicate a reason ("reden"); "aangezien" cannot be used in sentences that clearly indicate a cause ("oorzaak"). "Omdat" can be used without such a limitation3; so it can be used correctly in both sentences anyway. Therefore, it is an interesting experiment to try to replace "because" not by "omdat" but by "doordat" and "aangezien" successively. In A, "because" can be replaced by "doordat" but not by "aangezien", for "aangezien" suggests that there is an agent who considers what to do, but a thermostat cannot consider whether to turn the furnace on or not. It simply does. In B, it is the other way round: we can correctly replace "because" by "aangezien" but not by "doordat", for "doordat" makes one think of a duke acting like an automaton, for instance because the queen touches the threshold when she enters the room, and that by doing so she activates a mechanism that stabs a needle in the duke’s backside which makes him jump. But we cannot use "aangezien" in sentences that are apparently causal (i.n.s.) and "doordat" in sentences that are apparently reason giving, and this points to the fact that "because" in A indicates a causal (i.n.s.) relationship and that "because" in B indicates a reason giving relationship. My conclusion is that the relationship between the thermostat turning the furnace on and the temperature drop is different from the relationship between the duke standing up and the entrance of the queen, even though in English the same words are used for expressing these relationships. We have to make a distinction between relationships that are causal and relationships that are reason giving, and accordingly between two kinds of because, namely because I and because II.

2.2. I have shown that there is a distinction between reason giving and causal concepts and relationships and that a reason is not a kind of cause (i.n.s.). What the essence of this distinction is has not been discussed. Therefore, I now want to go more deeply into Dretske’s theory of representational systems and the role of beliefs for explaining behaviour by examining the third agreement of Dretske’s between the behaviour of the thermostat and the behaviour of the duke: that in both cases there is an internal state representing an external state, and that the duke as well as the thermostat do what they do because of this representation. But what does it mean that a certain internal state represents a certain external state, and that a certain kind of behaviour takes place because of this representation? In order to explain that, I want to have a further look at the explanatory role of beliefs in Dretske’s theory.

Above we have seen that, according to Dretske, cognitive factors (beliefs) are indicators of states external to an organism showing a certain kind of behaviour. In case a certain movement M of an organism is rewarded if state F is present and a certain belief C is an indicator of F, C will be recruited by that organism as a cause of M. C acquires, thereby, the function of indicating F and hence of representing F. Because C represents F, C can be a cause of M, and from this fact C derives its meaning, i.e. what it is supposed to say or indicate about what it represents. By having this meaning, C acquires also its relevance for explaining behaviour. That not F but C as a representation of F is a cause of M becomes clear, for instance, from the fact that C can misrepresent F and still can be a cause of M. However, we have already seen that the foregoing is only a part of Dretske’s causal theory of representation, for M will not take place without the presence of a conative condition D ("desire"). For example, when a rat sees food (C), it will not eat (M) as a rule, unless it is hungry (D). Hence also D is a cause of M, and also D has explanatory relevance for M (Dretske 1988, 79-131; 1990).

Another main point in Dretske’s representational theory is his distinction between symbols and signs or natural signs. Both symbols and signs have indicator functions. The difference is that indicator functions of symbols are assigned by us, while signs indicate in a natural way, having an objective relationship to what they indicate. A symbol may be a cross on a blackboard indicating a member of his team, when the coach explains the tactics to the players, while a small circle indicates an opponent. The coach might have done it the other way round, for the relation between a symbol and what it indicates is not intrinsic. Natural signs may be tracks in the snow or fingerprints: they indicate a fox or a thief independent of us, even if we do not see them, because they are objectively related to what they signify (Dretske 1988, 52-54).

Dretske uses the distinction between symbols and natural signs for characterising three types of representational systems. A representational system (RS) is "any system whose function it is to indicate how things stand with respect to some other object, condition, or magnitude" (id., 52). If the function of RS is to indicate whether O is in condition A or B, for instance, and RS has two indicators, a and b, for performing this function, then a and b are the elements representing about O that it is either A or B (as is the case). Depending on how a system performs its indicative and representational functions, Dretske distinguishes three types of representational systems. In an RS of Type I the elements do not represent intrinsically but the representational function is assigned by us, the makers of this RS: the elements of an RS I are symbols. Instances are maps, diagrams, gestures, codes etc. The way a map represents is conventional, and usually a key to symbols can be found in a corner of the map (id. 53-54).

Elements of RSs of Type II, on the other hand, are signs. The way they indicate is not conventional but natural. The way that signs represent may be conventional, however, because they can indicate several states at the same time. An electrically operated fuel gauge does not indicate only how much gas is left in the tank but also the downward force on the bolts holding the tank to the car’s frame and the amount of current flowing in the wires connecting it to the tank etc. As a rule, we are interested only in one function of the gauge: that it indicates the amount of fuel left in the tank; and this is what the gauge is supposed to indicate. This is not necessarily so, and because an RS II can indicate several states, what it has the function of indicating, i.e. what it represents, is determined by us. But because the way an RS II indicates is natural, the number of possible functions is limited, and this is why what an RS II represents is only partly conventional (id. 59-62).

The indicator elements of an RS may represent a state in a natural way, too. This is typical of natural representational systems (Type III). RSs of this type have their own intrinsic indicator functions, which are used by the system itself of which they are a part. This makes it difficult to give an example of an RS III, for how does one know that an element has the function to represent? In case of a fuel gauge, an RS II, we give it the function to indicate the amount of fuel in the tank. But how do we know that it is the function of a frog’s neural detectors to indicate the presence of edible bugs and not everything that flies by, including the small stones thrown by us in the direction of the frog? Or that a kind of marine bacteria has internal magnets, magnetosomes, with the function of indicating the oxygen-poor environment near the bottom of the sea and not the geomagnetic north? But if this is so, then we have two instances of an RS III. In the case of the frog, we can produce shadows simulating edible bugs. Then the "bug detectors" will no longer indicate the presence of edible bugs like in natural circumstances. Supposing that it is the function of the neural detector to indicate the presence of those bugs, shadows are misrepresented as edible bugs. As for the marine bacteria, in the Northern Hemisphere their magnetosomes indicate the oxygen-poor environment that the bacteria need to survive. It is therefore reasonable to suppose (but Dretske admits that it is controversial) that it is the function of these magnetosomes to indicate this environment; oriented by them, the bacteria propel themselves to the bottom of the sea. But if we put the bacteria into the Southern Hemisphere, they will swim in the wrong direction, to the surface of the sea, where they will die. The magnetosomes no longer represent the oxygen-poor environment and misrepresentation occurs (id., 62-70).

Against this background, we can examine Dretske’s theory how reasons explain behaviour. I shall limit myself to the main points and especially to the explanatory role of beliefs. I shall leave the role of desires aside, but my criticism can easily be extended to that part of Dretske’s view, too.

The essence of what Dretske wants to show is what the causal role of meaning is (id., 80), or as he says it himself "to understand how something’s having meaning could itself have a physical effect" (id., 83). In doing so, he demarcates his field: "Meaning" has to play a relevant role in the explanation (cf. the example of the soprano); for reasons, this means that we want to know the relevance of reasons as reasons. However, Dretske does not want to show the explanatory role of meanings as such: meanings themselves can have no explanatory role, according to Dretske, for reasons are not spatio-temporal events that can cause events. Hence Dretske’s conclusion: "in exploring the possibility of a causal role for meaning one is exploring the possibility, not of the meaning itself being a cause, but of a thing’s having meaning being a cause or of the fact that something has meaning being a causally relevant fact about the thing" (id., 80; italics by Dretske). In the last quotation but one, we see also that Dretske wants to know how having a meaning can bring about a certain effect. For Dretske, understanding the causal role of meaning is understanding the mechanism of causal working; the question what we believe is not a part of an explanation what we do.

In order to make clear how in an RS III reasons explain behaviour, Dretske first examines the case of an RS II like a thermostat (vid. figure 1; the letters have the same meanings as in the text).

Because belief C indicates and represents the external state F, figure 1 depicts how C causes and explains movement M (id., 84).

How does C get its indicator function? In the case of a thermostat, C is, let us suppose, a bimetallic strip given the job to make a current flow if the room temperature drops to 68oF (=F), as a result of which the furnace ignites (=M). We can also say that it is the meaning of C to indicate the temperature, and the fact that it indicates the temperature explains why C causes that the furnace is turned on. Conversely, the fact that C causes that the furnace is turned on, because C has a certain meaning, makes that C really has an indicator function of what it indicates (i.e. the temperature) and that C really has the status of a representation. As worded by Dretske: "An internal indicator acquires genuine (albeit derived) meaning … by having its natural meaning, the fact that it indicates F, determine its causal role in the production of output" (id., 87). However, being an RS II, the thermostat does what it does (indicating, representing, causing that the furnace turns on), because we have made it that way, and that is what makes it different from an RS III that does what it does in a natural way. Nevertheless, Dretske thinks that the structure of a thermostat as depicted in figure 1 is the prototype of both an RS II and an RS III (id., 88-89). If so, we can use figure 1 for analysing why the dukes stood up when the queen entered, since we are dealing here with an instance of an RS III. If we fill in figure 1 with the relevant data of the quotation of § 2, we get figure 2.

At first glance, we appear to have a simple model for explaining the duke’s action. If we can add a relevant desire for the duke in the scheme, we have a representation-theoretical version of the practical syllogism, or of Davidson’s primary reason followed by an action (cf. Dretske 1988, 109; Davidson 1989). We can regard Dretske’s approach also as a representation-theoretical formulation of the thesis – widely supported within the analytical theory of action – that there is no fundamental distinction between causal explanations of events and intentional explanations of actions, or rather that "Verstehen" is nothing but a kind of explanation (cf. bij de Weg 1996). However, by using the structure of RS II for expounding the structure of RS III, Dretske ignores (also according to himself!) the main distinction between Type II and Type III representations, namely that what C represents in an RS II is externally determined, while in an RS III, it is the system itself that determines what is represented. This becomes problematical, if what is represented by an RS no longer are objective facts but subjective facts, e.g. when the model of an RS III (and so in fact an RS II) is used for explaining actions.

In order to make this clear, I return to Dretske’s distinction between symbols and natural signs. Symbols get their indicator functions, because we have assigned them. In that sense, they are subjective (or intersubjective at most). They are the indicative elements of an RS I. Natural signs are "objective" in their relations to what they indicate: these relations exist independent of us. They are the indicative elements in RSs II and RSs III. But take the cases of a pigeon making pick movements or a rat pressing a bar, because that yields them food, examples used by Dretske for explaining his RS III. What is food? One can say that it is everything that is edible, and, as a rule, many animals consider everything to be food that is not inedible for them. What food is, is an objective fact, and as such it can be represented in an RS III by means of natural signs.4 But how about men?5 Maybe we can frame an RS III in which food is represented by means of natural signs in order to explain an action of eating, or anyway a certain kind of eating behaviour. The problem is that what is edible needs not to be food, for what food is is not only dependent on its physical composition and the constitution of the human body but also on the culture one belongs to. What is a delicacy in one culture may be considered inedible or unappetising in another one, because subjectively it is not food (tripe, dog meat). Or otherwise, what is eaten is a matter of status (caviar). Hence, it appears that what is considered as edible is symbolic in a high degree. Food symbolises status, civilisation, good taste etc. Nevertheless, what is considered as edible is limited by the fact that it must be edible, and that can be determined in a more or less objective way. So, there seems to be some room left to explain at least eating behaviour – but also eating actions? – by means of an objective relationship between natural signs and food as what is signified.

I shall not investigate whether this is really possible or whether it has any consequences for the representational model. But the case of the duke and the queen is different. In figure 2, I have reconstructed it as an RS III. In the RS II, depicting how a thermostat works, the bimetallic strip (C) indicates – and represents, because we have determined it that way – the room temperature. Analogously in the RS III of figure 2, there is a natural-sign relationship between our belief that the queen enters the room and the entrance of the queen (with our belief not only indicating the entrance of the queen in a natural way but also representing it "internally" – and that not simply because somebody externally has determined it that way). Or rather, this is the picture that is presented by Dretske. The difference with the RS II of the thermostat or an RS III in which food is represented as edible is, however, that what is indicated in the case of the duke and the queen is not an objectively existing fact, state or object (like the temperature, the edible), but it is a property that has been assigned to the queen. A woman is a queen, because other people have installed her as a queen or have given her this title because she has married a king. In the same way, she can be deprived of this dignity or title. This can happen voluntarily if the queen abdicates or if she divorces. But she can also be deposed. Rebellious subjects can proclaim the republic and appoint a president, or they can replace the royal house by another legal royal house (legal for them). Other subjects can refuse to recognise the change of power. Therefore, being a queen is for a woman not an intrinsic, or at least objective, property. In the same way that a symbol has certain properties because we have given it these properties, the queen is a queen because her subjects have recognised her: the queen is a queen for her subjects. Moreover, the recognition has to be affirmed continuously: if nobody treats her as a queen, she is not a queen (cf. bij de Weg 1996, 57-58).

Here it is not the point whether F (the entrance of the queen) or an element of F (the queen) is symbolic, or whether F is an subjective fact – or maybe an intersubjective fact –, but the point is whether what the duke does (rising to his feet) can be explained by means of an RS III like figure 2. For the duke may react to the queen’s entrance as if it is an objective fact. The point is whether it is possible to explain actions by means of an RS III, even if these are reactions to subjective situations and events.6 This seems to be the case, for instance, when we look at another action of the duke and let him function as a "thermostat" himself. Suppose that the room temperature has dropped to 68o (=F), that the duke starts to shiver, and that he turns the furnace on.7 Then we can explain, or quasi explain8, what the duke does by means of Dretske’s representational theory, and we can frame a scheme like the figures 1 and 2. Whether an RS III or maybe an RS II is involved, is not important, but what we have here is the objective situation that the temperature drops (F), that it is perceived by the duke (indication; he starts to shiver), and that he gets the belief that he feels cold (C); therefore he turns the furnace on (cf. Dretske 1988, 84).

At first sight, the case of the duke meeting the queen seems to pass off in the same way: the queen enters the room (F), the duke perceives it (indication), he gets the belief that the queen enters the room (C), and he stands up (M). There is, however, an important difference with the foregoing case. The temperature drop is an objective fact. Temperatures rise or drop also when nobody perceives it and nobody feels warm or cold (not to be confused with the fact that one can influence the temperature; the temperature change itself is objective). But as just explained, that the entering woman is a queen is not an objective, intrinsic property of hers. Whether the woman is a queen for the duke (or not) is so, because he recognises her as such (or not), or at least because other people do, and because of the subjective property that the entering woman is a queen for him, the duke honours her by standing up. If nobody (including the duke) would think that the woman is a queen, then it can happen that she enters the room and that the duke rises to his feet, but that it is not the case that he does so it because it is the queen who enters.

That being a queen is a subjective fact becomes even clearer, when we suppose that she just has been deposed. She is flying. Does the duke know that when the woman enters? Suppose he does not. Does he stand up then for the queen or for "just" a woman, a former queen? For the duke, she may be a queen but not for those who have deposed her. Or suppose that the duke has heard about the rebellion. The woman enters and the duke hesitates: if he stands up, he recognises her as a queen; if he pays his respects in another way, in fact he joins the rebellion. By doing what he does, by his action, he shows and determines what the woman is for him: a queen or not a queen.9 The situation of a rebellion is a marginal case, but it shows what normally does not come to light: the actions that we do "because" we believe that a certain situation (state) F exists may be a "consequence" of F, but they also affirm and constitute the same situation as being the situation F. F is "recursive" in the sense of Giddens (cf. bij de Weg 1996, 57). F both goes into the action as the situation that the agent reacts to, and it is the result of that action. F is input and output at the same time, or, in a certain sense, it is also the cause and the effect of what the agent does.

It is also because of this recursivity, that the structure of the case of the entering queen on the one hand and the structures of the cases of the thermostat or the duke reacting to the drop in the room temperature on the other are different. In the latter cases, the temperature drop has a certain meaning, namely that it becomes colder, and this meaning is preserved, even if the thermostat does not turn the furnace on because of a defect or if the duke fails to turn it on for some reason. The meaning of F before an action takes place is the same as its meaning after an action has taken place. We can call this meaning "single", analogous to the case of double hermeneutics. Then, the subjective meaning of the entering queen is "double". The queen has, as a "queen", a certain meaning for the duke, based on his past experiences: in the past, she was considered to be a queen by him and by other people, and she was honoured as such. From this point of view, the woman really is a queen. She looses this meaning for the duke, however, if he does not stand up for her or if he does not affirm this meaning in another way. Only if the duke rises to his feet (or does another affirmative action), we can say that it is for him the queen who enters. As a consequence, after the duke’s action, the meaning that the woman is a queen is fundamentally different from the meaning before the action: the meaning after the action is the reaffirmed meaning before the action; or it is, for instance, the meaning of a deposed queen or a queen that has abdicated voluntarily. Hence, the meaning that has to be affirmed by actions is double: there is always a meaning of F before and after an action. The meaning of F before the action has essentially already the possibility to be different after the action, the meaning after the action always is a little bit different from the meaning before the action, for it is reaffirmed meaning anyway. If the meaning of F after the action is different from the reaffirmed preceding meaning, then it still bears the "remembrance" of the preceding meaning. In this sense, subjective meaning is always "double".

What are the consequences of all this for the structure of Dretske’s RS III? In an RS II or an RS III, C has the function to indicate and to represent F. If the thermostat or the duke turns the furnace on, C indicates and represents the room temperature. But does C also indicate and represent that the queen enters in the case of the duke rising to his feet? Dretske answers this question affirmatively (explicitly in the quotation of § 2). As I have explained, however, being a queen is not an intrinsic but an assigned property. Hence, it is problematical to say that C is the belief that the queen enters and that it represents F, the entrance of the queen. This is best to be seen in marginal cases. In a marginal case, e.g. when the duke hesitates what to do when the queen enters, it becomes apparent that what "normally" seems to be the case, namely that C has the function to indicate and represent, simply is not so. Maybe C stands for a past experience, but it needs not to be so that the entering woman still is a queen. For knowing that, we need other evidence. The duke may have been informed about the rebellion, but it has been suppressed. Or the evidence can be "negative": the duke has no reason to belief that the entering woman is not the queen. He believes then that the woman that was a queen in the past at present still is, and so he stands up. What then happens is that on the basis of this belief the status of a queen is assigned to the entering woman instead of that this belief indicates and represents the queen. But then we can no longer say that "the queen’s entry into the room caused in [the duke] a belief that the queen was entering the room" (vid. the quotation of § 2). Just the opposite is rather the case.

Let us now look at the relation between C and M. In the case of the thermostat, C causes M. I do not go into the case of the duke turning the furnace on himself, but what about the duke’s belief that the queen enters? Does it cause that the duke stands up? We have seen that the duke’s standing up is not only a consequence of C, but that M, the duke’s standing up, is also an affirmation of the fact that it is the queen who enters and that this fact is constituted or reconstituted by the duke’s action. This action is not a simple physical movement; it is a mark of honour by the duke for the queen. However, the duke will perform this action only, if he really has the belief that the queen enters, unless he wants to feign that belief. If the duke feigns, then his standing up cannot be caused by this belief but only by another one plus the reasons why he wants to feign at most. But even if the duke acts on the basis of the belief that it is the queen who enters, we cannot maintain that this belief causes the fact that the duke stands up, for that the duke stands up is an affirmation for him that it is the queen who enters (and not a deposed queen or just a woman). It is a part of a complex of views that make that it is the queen who enters (and nobody else). Of course, the action of standing up as such is not important, and the duke could have chosen another action for honouring the queen. But for any action within an interpretative frame it is so that this action is not only a result from this frame, but it is also constitutive of the frame of which it is a result.

When Dretske puts the explanation why the thermostat turns the furnace on on a par with the explanation why the duke stands up for the queen, then what he in fact does is confusing physical facts and relationships and social facts and relationships. A physical fact gets its meaning for a person only if this person is confronted with it. Of course, this is also true for a social fact, but what makes it different from a physical fact is that a social fact has already a meaning (given by other people) before the person involved comes into touch with it, and this person can say that he has understood the meaning only by defining his position towards it by affirming or denying the meaning "offered" to him.10 I shall not go into the question whether so-called "physical facts" are not actually social facts for an acting person in the most cases. I think that it is so. The point is that Dretske makes no fundamental difference between an indisputably social fact like the duke’s standing up for the queen and a physical fact like what the thermostat does. It is true that a thermostat is made by men, but a thermostat functions also in case the engineer who has made it has died and in case there are no people to use the apparatus or who know what it is, as long it is intact. Physical phenomena continue to occur despite of the meanings that we have given them. Social phenomena cannot exist as such without meanings assigned by the performers or observers of them.

My analysis cannot be without implications for Dretske’s representational theory and model, since he takes no notice of the idea of double meaning and that people act on it. Moreover, we have seen that the "first" meaning of a social phenomenon is not intrinsic but assigned. We may call this signification "symbolical", but for Dretske a symbol and what is symbolised are not identical: the cross on the blackboard represents the player but it is not the player. The woman, however, does not represent the queen: she is the queen. Does she represent the kingship? But what could that kingship be without the queen? Therefore, neither Dretske’s RS II, nor his RS III (that in the way it represents is not different from RS II), nor his RS I are models that can be used for explaining action ("behaviour" in Dretske’s terminology) based on a double meaning. But if we accept that double meanings are fundamental for actions, then the question how to explain an action from the meaning given to it by the agent is no longer a matter of how to explain it causally but how to understand ("verstehen") it (bij de Weg 1996).

Above, I have made a distinction between a causal "because" (because I) and a reason giving "because" (because II). The thermostat turning the furnace on exemplified the former, the duke standing up for the queen the latter. I have made clear that the first relationship can adequately be described by Dretske’s representational model but that the second one cannot, because it does not include the idea of double interpretation. It is not up to me to elaborate how an interpretative scheme (analogous to figure 1) would look like. But if "cause" (i.n.s.) and "reason" have corresponding functions, although they indicate different relationships, then it is obvious that what in the representational model is a causal relationship is a reason giving relationship in the interpretative model. In the representational model, C causes M on the basis of the fact that it is an indicator of F. C is a belief: either the "belief" of the thermostat that the temperature drops, or the duke’s belief that the queen enters. So, the latter belief has the same position as the former. However, we have just seen that the duke’s belief cannot be the cause of the fact that he rises to his feet. It is the other way round: that the duke rises to his feet affirms that the entering woman is a queen. The "causal direction" is here not from C to M but from M to C. It is true that we have said that the duke stood up because the queen entered, but this "because" is not causal (i.n.s.) but reason giving. Whether we have to do with a reason or a cause (i.n.s.) depends on the role of interpretation. We talk about a reason for what an agent does in a social situation, in which C is both a precondition for that situation and a consequence of it, and in which C needs a subjective interpretation by the agent. C has a meaning assigned by the agent. The situation changes if the agent’s interpretation changes. We talk about a cause (i.n.s.) of what an agent does, if there is an objective (often physical) element that steers the agent despite of himself: the temperature drops anyway; the agent cannot change that by reinterpreting the situation. But the agent can ignore the temperature drop, and he does not need to turn the furnace on, even if he feels cold. Hence it is that the temperature drop steers what the agent does only "in principle", and this is why I have talked about a "quasi-explanation": also the temperature drop needs an interpretation by the agent and his reaction to it is not automatic. But if we really have to do with a cause, then the relevant meaning is not subjective (or intersubjective at most) but objective, representing a certain state and informing about it, and it is only this objective meaning that Dretske must have had in mind, when he framed his RS III for explaining what an agent does.

If a reason is not a special kind of cause but a determinant of actions sui generis, it cannot be so that a reason explanation is a kind of causal explanation (i.n.s). It is an explanation of its own kind. As I have argued in my 1996, a reason explanation starts from the perspective of the agent and uses this perspective to understand the meaning the agent himself gives to a situation and why he does what he does. A causal explanation starts from the perspective of the investigator and tries to track the causes of what the agent does from this point of view. The consequence is that we need two different explanatory models.

2.3. I want to conclude my analysis of Dretske’s referential theory with an investigation of points 2) and 4) of § 2 that refer to Dretske’s concept of cause. I shall use the upshot of the preceding section as my frame of analysis.

When we ask what the causes of a process are, we can answer this question in two ways, according to Dretske. We can either look for the event that triggers the process, and then Dretske speaks of the "triggering cause"; or we can look for the background conditions that made that this process has a certain form or structure, so that it is M1 and not M2 that is the consequence of C. Then, Dretske speaks of the "structuring cause". A temperature drop causing to occur certain events in the thermostat, while in turn these events cause the furnace to ignite, is an example of a triggering cause. The structuring cause is what makes that the thermostat turns the furnace on and does not open to garage door, when it becomes cold. And this can happen either because the thermostat is wired to the furnace in a certain way, or because the electrician wired it that way. So structuring causes can be of two kinds: "(1) the background conditions that enable the one thing to cause the other or (2) whatever earlier event or condition that brought about these background conditions" (Dretske 1988, 42). There is also a difference in time perspective between a triggering cause and a structuring cause. The former makes that the process takes place now; the latter concerns already existing relationships that have been made in the past. The distinction between triggering and structuring causes is important, for according to Dretske, reasons are structuring, not triggering causes (id., 37-50, 114-115).

This is not right, however, insofar as reasons and causes are fundamentally distinct concepts that refer to categorically different relationships. I do not want to dispute whether the term "reason" can be used better for Dretske’s structuring content-possessing mental states or for my "because II", but I now want to read the last sentence of the quotation of § 2 that way that it says that "reason" (in my sense) is a concept that refers to the presence of a certain structure and not to the incidence of an occasional event. But given the categorical distinction between "cause" and "reason", does there exist a difference between "triggering reasons" and "structuring reasons" in the same way as there exists between triggering and structuring causes?

For answering this question, I make use of my 1996. I distinguished there several factors that make an agent act. The main factors are the situation the agent finds himself in, his reasons, and his intention. The intention is what the agent acted for, or more or less what Dretske called "desire". For the sake of simplicity, I shall ignore it here. The agent’s reasons are his relevant past perceptions and experiences. They are the considerations – implicit and explicit – that move an agent to develop an intention and to act. The situation involves the circumstances that provoke the agent to develop an intention and to adduce reasons for it. What are then the reasons and the situation in the case of the duke rising to his feet? To make this clear I use another description of Dretske’s describing this instance:

There is a clear difference between explaining why, on the one hand, Clyde stood up then and explaining, on the other hand, why what he did then was stand up (why he stood up then). He stood up then because that was when the queen entered, or when he saw the queen enter, the room. He stood up then as a gesture of respect. The difference between citing the triggering cause of a process (the cause of the C which causes M) and what I have been calling its structuring cause (the cause of C’s causing M) reflects this difference."(Dretske 1988, 42-43; italics by Dretske)

What is Clyde’s situation? The italicised words mark our options: "then" and "stood up". However, "stood up" cannot denote the situation that provoked Clyde to act, for this is simply the action that we want to explain. What is meant by "then" is rather vague, for one does not react on everything present, but Dretske adds an explanation: "then was when the queen entered … the room" ("or when he saw the queen enter the room"; I shall leave this implied). Dretske himself calls it the triggering cause why Clyde stood up. Does this mean that I have to say that it is the triggering reason why Clyde stood up that must be distinguished from his structuring reason for acting? For answering this question, let us examine first what the structuring reason for Clyde’s action might be. The quotation makes clear that we are looking for an answer to the question why the duke (Clyde) stood up, and not – the examples are mine – why he did not shout "Long live the queen" or sing the national anthem. What is at stake are the duke’s reasons what he is in the habit of doing under certain circumstances. Dretske would speak then of the structuring cause of an action, and I have rejected that. However, this does not alter the fact that an agent like the duke will have at his disposal certain patterns of action that function as reasons that are used to determine what to do under certain circumstances (bij de Weg 1996, 287ff). The duke knows that on certain occasions the national anthem is sung and that on other occasions "Long live the queen" is shouted, and he knows when to do what. He knows, too, that in the case described, standing up in order to honour the queen is an appropriate action. Hence, the duke has a justification, and so a reason, to do so when the queen enters.

This does not imply that the duke really will stand up, when the queen enters. For he will rise to his feet at that moment only, if he wants to recognise her as queen and wants to do that publicly, or if he wants to feign that he recognises her. In other words, he will stand up only, if he wants to assign her the meaning of queen. If not, maybe he stays in his seat. Under normal conditions, such an assignment happens without thinking or even unconsciously, but in principle, it is not so that the duke "simply" acts, but he acts on the basis of certain considerations, and these may be explicit. Therefore, it is not obvious that the entrance of the queen brings the duke to his feet, because (i.b.s.) he has structural reasons (his individual patterns of action) for it. Thence, that the entering woman is the queen, is not a "then", an event that happens to take place at a certain moment. Much has preceded it, and the event as it is, just like the duke standing up for the queen, is founded on certain past perceptions and experiences.

But then, it has no sense to say there is a triggering reason that makes the duke standing up "then". What looks like a triggering reason in fact is related to a complex of considerations detached from the present. There simply is no such thing like a triggering reason that can trigger off a mechanism as a structuring reason in order to make the duke act and that can be distinguished fundamentally from this structuring reason. What Dretske has called a "triggering cause", and what here might be a "triggering reason", is itself part of the "mechanism" that determines whether an action will take place.

But of course, giving meaning does not happen just like that, and, of course, something happens at the place where the duke is. Reasons are not applied unreasonably; meaning is not given without an object. The duke is in the room with other guests. Everybody is waiting with bated breadth what will happen. The door is opened. A woman enters. Who is she? That is what I have called the situation: the circumstances that provoke to act. By this, the situation becomes the circumstances that are relevant for action and that mark a frame for action. In case the entering woman is simply another guest, most people present may only nod politely in greeting and continue talking. In case she is the queen, then everybody stands up and bowss. But the situation, an element in the situation, will never determine an action, even if certain structural requirements have been met, like the way a temperature drop can turn a furnace on. "Social" meaning is not automatic but it is assigned. This is the essence why there is no distinction between a triggering and a structuring reason, and why a duke is not a thermostat.


1 I have chosen just these two sentences, because here Gean’s confusion between "reason" and "cause" is most apparent. At this place, I do not want to discuss whether actions as such can have causes (cf. bij de Weg 1996, ch. VI) but only that reasons have to be distinguished from causes.

2 But can we then really say that the agent had "left" the party? "Leaving" is something an agent does, not what happens to him.

3 In fact, there are five causal connections, but "daar" is used only in written language and "vermits" is regional. I do not need them for my analysis. Some linguists deny that "omdat" can be used for indicating an "oorzaak". If we would accept that, it would make my point even stronger.

4 At least, let us suppose so, for if not, we can immediately move to the next case.

5 And maybe we must add: how about (certain) higher animals?

6 I shall not discuss here whether actions can be explained anyway, even if they are supposed to be reactions to "objective" situations and event. Vid. bij de Weg 1996.

7 Leaving aside that the duke finds the low temperature unpleasant. But as stated before, I give no notice to the agent’s desires. Moreover, it is not quite correct to say that the duke’s shivering is an indicator, for shivering itself is a kind of behaviour that is a consequence of the fact that the duke feels cold. But analogously to the situation that temperature fluctuations cause a bimetallic strip to expand or to shrink, for brevity’s sake I shall consider the shivering itself as an indicator.

8 Why I talk about "quasi explain", vid. § 2.2.9 

9 Of course, the duke can stay in his seat, because at that moment he prefers not to show openly that he recognises her as a queen. This does not affect my argumentation.

10 From a methodical point of view, this is the same as what Giddens has called "double hermeneutics". Vid. Giddens 1986, 120; cf. bij de Weg 1996, IV; Habermas 1982, 162ff.


Beckermann, A. (ed.) (1977), Analytische Handlungstheorie 2, Frankfurt am Main: Suhrkamp

Davidson, D. (1989), Essays on Actions and Events, Oxford: Clarendon

Dretske, F. (1988), Explaining Behavior, Cambridge, Mass. etc.: MIT

             Dretske, F.(1989), "Reasons and Causes", in Philosophy of Mind and Action Theory, 1-15

Dretske, F. (1990), "Précis of "Explaining behavior: Reasons in a world of causes", in Philosophy and Phenomenological Rresearch, 783-786

Gean, W.D. (1966), "Reasons and Causes", in The review of metaphysics, 667-688

Geerts, G., et.al. (1984), Algemene Nederlandse Spraakkunst, Groningen etc.: Wolters-Noordhoff, (= ANS)

Giddens, A. (1986), New rules of sociological method, Londen: Hutchinson

Habermas, J. (1982), Theorie des kommunikativen Handelns. Band I, Frankfurt am Main: Suhrkamp

Weg, H. bij de (1996), De betekenis van zin voor het begrijpen van hand, Kampen: Kok Agora