Empirical actions (and some other things)

This is mainly an (yet unsuccessful) attempt to develop what I understood as a Wittgensteinean suggestion here for how we could justify claims of the form ‘I know that p’, where p describes what is happening in the speaker’s environment, by performing certain empirical actions. Since these are my personal notes, some other thoughts are intermixed. If you have any feedback, please leave a comment bellow.


Two possible discourses:

“We have conscious mental states. This is what makes us different. This is perhaps essential for us being persons. Conscious mental states do exist, since they are reducible to brain states, which can be observed and studied scientifically.”

“We are able to perform actions. That is, we can assume responsibility for some events, under certain descriptions (in short). Since only a person can perform actions (thus defined), this is what makes us persons. This is, perhaps, what is special about us, although the distinction between persons and non-persons might, in fact, admit of degrees. Actions do, of course, exist. Whether they are reducible or not to physical (or natural) events might be disputed, but they can be observed and studied scientifically (this is possible because one can interfere with an action and even act upon another action – to observe an action is to empirically act upon it).”

I am inclined to prefer the second discourse to the first one. But are they exclusive? And if they aren’t, how are they related?


“But how can one be responsible for what one observes?”

One is not responsible for one’s sensory input, but when one is making an observation, one can choose what to observe.

“But you need sensory input in order to make observations.”

Analogy: the sensory input is the raw material; this is given; but I can chose what to to make out of it; my choices are limited by the material used, but I am responsible for my choices.


“Anyway, the regular problem (for empiricists) was to justify particular sentences about regular physical objects based on sentences about sensory input (protocol sentences). Now you seem to say that sentences about physical (or natural) objects are justified on the basis of our empirical actions, which are performed directly on objects from our environment. Let’s leave aside the fact that astronomical observations of distant stars might not be actions on existing objects (since some of them might have already disappeared). We could also leave aside for now the fact that some sentences which seem to be about physical objects are justified, actually, on the basis of sentences about our sensory input. It still seems to me that you have just pushed the problem one step back.

Actions performed on our environment have a conceptual structure only due to our acceptance of them under some description. So now, what justifies us to describe an action A as ‘grabbing a chair’? I suppose it has to be our sensory input. There is still a gap left between the two.”

There is no gap. Sensory input guides my action of grabbing a chair. All this has nothing to do with knowledge or justification, but with the workings of my body. Also, no actual description of an action is needed for it to have a conceptual structure.

“But then you have to say that the action has a non-linguistic conceptual structure, in the way in which a monkey grabbing a chair might be performing an action with a non-linguistic conceptual structure. So how would you justify ‘This is a chair.’ based on such an action?”

Well, when we talk about justification we are not interested in causal connections. My sensory input naturally enables me to perform actions like grabbing the chair. The action of grabbing the chair enables me to say that the object I am grabbing is a chair. The same action might not enable the monkey to form a propositional belief, but that is besides the point.

Grabbing a chair makes one responsible for accepting the opinion that the grabbed object is a chair. One might be unable to opine that the grabbed object is a chair, but still be responsible for doing so.

You might say that one should not have a duty responsibility towards something which one is unable to do. This seems plausible, but the opposite is not inconceivable.

In any case, on that ground one might deny that the action performed by a monkey (or by a baby not having learned to speak) has the same conceptual structure as the action performed by a talking human. The latter action has the conceptual structure of the description the agent would accept (or provide) for it.



“What would you reply to what Sellars says in ‘Language as Thought and as Communication’? The passage I have in mind is this:

<< […] it can scarcely be over-emphasized that to approach language in terms of the paradigm of action is to make a commitment which, if the concept of action is taken seriously, leads to (a) the Cartesian idea of linguistic episodes as essentially the sort of thing brought about by an agent whose conceptualizing is not linguistic; (b) an inability to understand the rule-governed character of this conceptualizing itself, as contrasted with its overt expression. For if thought is analogous to linguistic activity to the extent implied by Plato’s metaphor ‘dialogue in the soul,’ the idea that overt speech is action and its rules rules of action will generate the idea that all inner speech is action and its rules rules of action, which leads to paradox and absurdity without end. >>

It seems to follow from this that if you think of language as a practice consisting in linguistic actions only, you cannot solve the rule following problem raised by Wittgenstein.”

Let me start by distinguishing between the interlocutor (towards whom a speech act is directed) and the addressee of an action. When I am asking a person who is bothering you to stop, I am doing it for you. My interlocutor is the person to whom I am talking to, but the addressee of my linguistic action could be you. Now, what Sellars seems to say is that when one thinks-out-loud there is no interlocutor. But there could be an addressee. Thinking-out-loud could be an action, even if it is not the performance of any speech act. Sellars thinks that rules of criticism provide an escape from the RFC and assumes that something cannot be a rule of criticism if it regulates actions.

Now suppose that I encourage my son to express his wish to go outside by bringing his shoes to me. He doesn’t speak yet, but has learned a rudimentary form of communication. I suppose he has partial concepts for ‘shoes’, ‘bringing something to somebody’ etc. His bringing his shoes to me could be considered a rudimentary action. Of course, he is not able to accept it under this description, to assume a responsibility for it in the full sense. It might also be the case that it is not in his control to refrain from doing it. Also, he could not do it ‘as a joke’. Several things are missing, but I could still call this a rudimentary form of action. If this is the case, then such a rudimentary action can be regulated by a rule of criticism.

“But what is going to distinguish an action from something which is not an action, then? It seems now that you are ready to call all behaviour action. If that is the case, you might as well talk about all events as if they were actions and identify their causes as their agents.”

You miss my point. I do not think that one can solve the problems raised by Wittgenstein’s RFCs by talking of rules of criticism. What distinguishes rules of criticism from natural regularities? The fact that they are preserved from one generation to the other? Teaching someone a rule of criticism is not different from reproducing the functions of an artifact in a new artifact. Rules of criticism can only be syntactic. But if one wanted, one could instil a “syntax of actions” into somebody.

The rule following problems do not appear because we conceive language as a practice, made up of actions. They only appear when we try to theorize about the practice in case. This is true for any rule governed practice.

Sellars seems to say the following:

(i) Not all rule-governed practices need to have a conceptual content, but if language is to have some conceptual content, it has to be a rule-governed practice.
(ii) If anything linguistic is an action, then the linguistic rules are rules for action.
(iii) In order to follow a rule of action one must understand the rule.
(iv) So the linguistic rules of action must also have a conceptual content.
(v) Then, the rules must be formulated in something like a language having conceptual content.
(vi) Since the rules must be understood for a language to have conceptual content, the rules must be independent of the spoken language, so the language in which the rules are expressed must be different from a spoken language (something like a language of thought).
(vi) The language of thought in which the rules are expressed must be governed by rules in order to have a conceptual content.
(vii) Those second order rules themselves must have a conceptual content as well.
(viii) Here we have an infinite regress.
(ix) Therefore, not anything linguistic is an action.

Rules of criticism are such that one can follow them without understanding them. But, then, why not reject (iii) instead of (ii)? Why not say that there are rules of action such that one can follow them without understanding them? Why not say that understanding a rule is different from understanding a sentence?

After all, to understand a rule is to take part to a practice. To understand the concept “rule” is to be able to distinguish between rule-governed practices and activities which are only limited naturally. There has to be a gradation between the two, of course. Thus, the concept could not receive a complete analysis (and things are the same in the case of other concepts: “analytic”, “person”, “mind”, etc.). However, this does not mean that the concept cannot have its uses.

In any case, I do not aim to offer a theory of empirical justification here (or a theory of the rational relations between communicative and non-communicative actions elsewhere). My aim is different. I am trying to produce a conceptual proposal. The proposal could be accepted if it proves useful. Its use is to avoid several philosophical problems which appear when we try to conceptualize the relation between our propositional knowledge and our experience (or the relation between our language and our environment; or the fact that language has conceptual content).

Let me list a few points which I think we do not want to give up:
(1) Most of our propositional knowledge is based on our experience. That is, we can justify (at least) what we believe about our environment by our experience of our environment. (Giving this up would amount to saying that the brain in the vat and similar scenarios are plausible.)
(2) A belief is justified only if there are reasons for it. One can naturalize reason, but in order to talk about justification one has to be able to spell out the rational relations between a belief and its grounds (what justifies it). Even if logical implications are thought of as causal relations between beliefs, you cannot start directly with causal relations between beliefs and talk about justification. A belief may lead to another by some other route. (Giving this point up would amount to say that our beliefs are the product of our environment, but there is no justification for holding them.)
(3) Nothing can follow by reason from something which does not have a conceptual content. (If we give this up we are going to give up our concept of reason.) My belief that I have such and such a sensory input might be produced by my sensory input, but it does not follow by reason from my sensory input. (If it did, the same belief could follow by reason from different sensory inputs, incompatible with each other. This could be true in logic, where, if A is true, then B -> A and ~B -> A are also true, so A could follow both from B and ~B. But this would only mean that a true belief does not need justification. Its truth would make it possible for the belief in case to be “justified” by anything. However, this can only mean that ‘follows by reason’ does not mean ‘follows by logic’.)

Here we have the philosophical problem noticed by Sellars and McDowell. If we conceive our experience as some sort of raw sensory input, we can justify nothing from it (according to 3). But a least our beliefs about our environment must be justified by our experience (according to 1) and so our beliefs must follow by reason from our experience (according to 2).

One solution is to conceive our experience as different from a raw sensory input. McDowell does this when talks about our sensations having a conceptual content. This proved to be misleading in several ways. Also, sensations with a conceptual content can still be illusory (in the same way in which raw sensory input can be illusory). The gray dots appearing in the grid illusion do not exist. It does not matter that I see them as gray dots. But I can not observe them, circle them, measure them a.s.o. (also, I should say that observing something is not, simply, focusing your attention onto some sensory input; I might be able to focus my attention on a particular gray dot without looking directly at the place in which it appears to be located, but by that I would not be observing a grey dot).

This is why i want to talk about experience as the performance of empirical actions directly on our environment. One could try to observe something and fail, but the action of observing a gray dot on a sheet of paper (under this description) would justify the belief that there is a gray dot on that sheet of paper. One could have the belief and choose not perform the action, but justify that belief in some roundabout way. One could perhaps observe the gray dot and forget about it, thus being unable to attribute the respective belief to oneself. Several other cases could be imagined. The action itself does not have to be justified. It only has to be successful and correctly described. The second condition introduces a bit of holism into the whole account, perhaps. The relation between the two actions – the empirical non-communicative one and the belief attribution (even if one conceives it as being done in a thinking-out-loud or thinking-for-yourself way) – is that of an action following by reason from another. This could also be conceptualized by saying that the (successful) performance of the first makes one (duty) responsible for the performance of the second. (Or, if you wish, the performance of the first produces a duty responsibility for the performance of the other, no matter the agent, in the same way in which the opening of a locked door produces a responsibility for unlocking the same door).

“But, while one cannot open a locked door if the door is not unlocked, one can say that one believes that there is a chair around without anyone observing the chair around.”

One could, of course, say it. There are no physical limitations and impossibilities when it comes to saying (or believing) things. One cannot open locked doors which were not unlocked first, but one can hold unjustified beliefs. However, one cannot hold a justified belief which had no justification. So it could be said, perhaps, that the attribution of the justified belief that there is a chair in this room is not possible without the performance of an empirical action on the chair in this room.

“But then you need the concept of a ‘justified belief’ in order to say what it is for an action to follow by reason from another one (at least in this case).”

Ok, let me try again.

The opening of a locked door produces a responsibility towards the unlocking of that door. I use “to produce” here, but the relation is not causal.

In the same way, if I move a piece to check your king in a game of chess, I must be responsible for bringing that piece in the position I started my move from. For instance, if the move was e5-6+, then I am responsible for bringing the pawn to e5 by previous moves.

I might have some reasons for attributing to myself a belief I have no justification for. An implicit attribution of belief that p to myself could be done by opining that p. When I say ‘I opine that p’ or ‘I think that p’ or something similar, instead of simply stating that p, why do i do it? Perhaps this is because something is holding me back. Otherwise I could simply assert that p. Opining that p seems to be weaker than asserting that p in this respect. Nevertheless, it could be considered that believing that p is a necessary condition for asserting that p.

Now, suppose I say ‘There is a chair in this room’. (Maybe I could just think-out-loud ‘There is a chair in this room’, I don’t know. In any case, I cannot think of myself saying ‘I think that there is a chair in this room.’ Some weird circumstances should be imagined for that to happen naturally and I cannot imagine them right now.)

What is the relation between this action and an empirical action performed on the chair in this room? I have said that the first action follows by reason from the second. That is, if I am responsible for having said that there is a chair in this room then I am (or at least somebody is) responsible for observing the chair (in some way or another). Why is it so?

What if I said ‘There is a pink elephant in this room’? If nobody checked whether it was a pink elephant in the room… I mean, what I said either was a move in a language game or it wasn’t. If the game was to say whatever crossed our minds and ignore it, then fine. But in the usual game, if something is said, one can agree or disagree. If one agreed with whatever I said without checking, then one would have contributed nothing to the game. Now, if one checked whether there was a pink elephant in the room or not and replied that I am wrong, what would I do? I could accept that I was wrong or not. If I did not accept that I was wrong, then I would have to prove (by some empirical actions) that I was right. Otherwise I would have to accept that I was wrong. Doesn’t that mean that my ‘expression of belief’ produced a responsibility with respect to an empirical action?

If one said that by thinking-out-loud one played no language game, this could be rejected in such a case. But then I cannot see how one can express beliefs by thinking out loud.

“A door is unlocked in order to be opened, but one could unlock the door and never open it. This is way we say (actually, this is why you said) that the opening of the locked door does not follow by reason from its unlocking, but its the other way around. It seems to be the other way around precisely because one cannot open a locked door without unlocking it first. Stating the conclusion of an argument makes one responsible for stating the premises, but not for stating whatever follows from that argument. You might have stated the conclusion of that argument in order to use it as a premise in another argument, but you do not have any duty to produce that other argument. There must be something wrong with what you think.”

Ok, let me try to say this one more time.

I say that all men are mortal and that Socrates is a man. By doing this, I have a duty to accept that Socrates is a mortal. In this sense, ‘following by reason’ relations transfer responsibility from one action to another. In the case of non-communicative actions, however, things usually happen as if I started by saying that Socrates was a mortal and provided the reasons afterwards. As if there would not be possible to provide those reasons without saying what they were reasons for. This was the case with unlocking the door and opening it. Unlocking the door follows by reason from opening a (previously locked) door. Thus, the action of opening the door produces a responsibility for performing the act of unlocking the door.

Now, on this model, it should be the same in the case of empirical actions. Observing the chair in my room makes me responsible for believing (opining or attributing to myself the belief) that there is a chair in my room. If we model this on the case of non-communicative actions, it should be as if I state my belief first and then observe the chair (I first unlock the door and then open it). So it should be as if I could not observe the chair without believing that there is a chair in my room. Here is were I went wrong at first (actually, you misled me). It is clear now that if consider the action of observing the chair in my room under that description, one would not describe what one does like this if one did not believe that there was a chair in the room.

“I am but an imaginary interlocutor, so there is not difference between me misleading you and you getting something wrong.”

That is true, but when I am writing your lines I am not paying as much attention to my view as I do when I am writing my lines (since you are only objecting to my view and not presenting and developing it).

“In any case, it seems now that in order to empirically justify the belief that there is a chair in your room you first need to believe (without any justification) that there is a chair in your room and then empirically act on the chair in your room and succeed to do so. Doesn’t this seem weird to you?”

Not at all. In fact, I only need to observe the chair in my room. The belief that there is a chair is, if you wish, a presupposition of my action (as it is when I sit in the chair in my room). I do not need to say or think that there is a chair in my room in order to sit in it. There is nothing psychological here. I am responsible for holding that belief, but no psychological act of ‘entertaining that belief’ is needed. This case is similar to the one of unconscious beliefs. My responsibility for holding that belief amounts to my acceptance to attribute it to myself for the past. I must agree that when I sat into that chair (if that is what I accept I did) I had the belief that there was a chair there.

“But that would have been an unjustified belief, wouldn’t it?”

When I sat in the chair it was justified. The same holds for observing the chair. You might insist that I had the unjustified belief while I was reaching towards the chair or something like that, but I could deny that reaching towards the chair was something I did.

To conclude:

‘A follows by reason from B’ means that:
(1) doing B is a sufficient condition for the (duty) responsibility for doing A.
(2) the (duty) responsibility for doing A is a necessary condition for doing B.

This can be applied for the case when A is the action of explicitly attributing to oneself the belief that p or the action of opining that p (or even thinking-out-loud – or for-oneself – that p in a game which calls for agreement or disagreement; in short, A is the action of believing that p) and B is a (successful) empirical action performed directly on the object(s) p is about.

Thus, a belief can be justified by an empirical action.

“But what if you pretend to observe a pink elephant? How can I make sure that your empirical action performed to justify your belief is not pretended?”

Well, I suppose that the empirical actions by which we can justify beliefs to each other can be performed in cooperation (we can observe the chair together, but we cannot observe the pink elephant together).

After all, if reference is assured by cooperative tagging performed directly on objects from our environment, why not accept that shared justification consists in cooperative empirical actions? It would appear this way that the two are linked. For instance if a tag an object which I hold in my hand (and you can observe) as food to you, we are both justified in believing that the object does exist. And we can say that the tagging was successful in this respect, since it was the tagging of an object justifiably believed to exist, although it might prove wrong that it was food.

“Can it be said, then, that tagging x as food makes you responsible for believing that x exists?”

You are trying to mislead me again. It is not that tagging x is a sufficient condition for being responsible for believing that x exists. Rather, being responsible for believing that x exists is a necessary condition for tagging x.

But this is true only if we assume that a responsibility for observing x is a necessary condition for tagging x. It is disputable, however, that the action of observing x follows by reason from the action of tagging x in some way, since the action of observing x is not a sufficient condition for being responsible for tagging x in some way.



“Acceptance of some presuppositions (there is a present King of France) is a necessary condition for making some assertions (The present King of France is bald). However, the relation between the content of an assertion and that of one of its presuppositions is not one of logical consequence.”

Asserting that there is no King of France at present can be, in some sense, opposed to asserting that the present King of France is bald. Given their conceptual content, the two are opposing actions. Logical contradiction is a species of such an opposition. From this it follows that “proper logical consequence” is a species of a more largely conceived “rational consequence”. For this more general relation (between actions) the model was already outlined:

‘A follows by reason from B’ means that:
(1) doing B is a sufficient condition for the (duty) responsibility for doing A.
(2) the (duty) responsibility for doing A is a necessary condition for doing B.

According to this model, then, accepting the existence of the present King of France follows by reason from asserting (in a non-fictional context etc.) that the present King of France is in some way or another. This is so because it is enough to non-fictionally assert something about an object to have a responsibility to accept the existence of that object and also because you cannot assert something about an object unless you assume responsibility for accepting that the object in case does exist.

Now, according to a logical rule, from the assertion that some particular object has the property F it follows that there is at least an object (indefinite) having F. This could, of course, be presented in the same way: accepting that a property F is satisfied (or instantiated) by at least an object follows by reason from (non-fictionally) attributing that property to a particular object in an assertion (or in tagging). The justification is similar: the attribution of F to an object is enough for one to be responsible for accepting that at least an object has F and one cannot assert that an object has F unless one assumes a responsibility for accepting that F is instantiated.

If a distinction between the two cases must be provided,…


“Humans are mortal.” – Here the logician says that it is part of what is said that humans do exist (or at least one of them). So of course it follows logically from what is said that humans do exist.

“Socrates is mortal.” – Here things are not as clear (see the Russell – Strawson controversy).

“No goblin is human.” – there is no x, such that… – from this does not follow that there are either goblins, or humans.

One reaction is to say that the quantifiers do not actually capture the concept of existence. They are symbolic means used primarily to confer clarity to the mathematical language. “The divisors of any odd number are odd” and “No divisor of an odd number is even” have no ontological commitment. Their logical form says nothing about their ontological commitment. It rather says something about the way in which one could produce a proof for one of those sentences (by mathematical induction or reductio, respectively).


Action, conceived as behaviour, does reach to the real world (our environment) in a way in which thought did not. The old problem of epistemology, in this respect, was this. Our knowledge of reality has a conceptual structure, but how can it be “of reality”? How does our thought (which is required for having concepts) reach to reality? How can the intentional object of a thought be a real thing?

We do not have such problems when we think of actions, of course. They reach to reality. Their intentional objects can be the objects acted upon, that is, the object causally affected by our behaviour.

But now the problem is: ‘How can human behaviour, which is composed of natural events, have a conceptual structure?’. The answer is that actions are caused by intentions, which have a conceptual content. But now we are back to square one, since we have to understand how can the intentional object of the intention accompanying an action coincide with the object acted upon.

Wittgenstein’s solution was understood (by Kripke, for instance) as saying that concepts are embedded into our practices – via our semantic rules. (I think Wittgenstein did not want to provide a theoretical solution to a theoretical problem, but to end the discussion.)

However, Wittgenstein himself was left with some interesting problems. Natural events, by themselves, have nothing to do with knowledge. Actions are different. But how can we say that they are caused by intentions? If the intention to move my hand is a psychological event, then (PLA aside) it must be caused by something. If the action is mine, then the intention must be mine, too. Perhaps I do not need to cause it in order to call it mine, but there has to be an intention causing event. And if that event is not under my control,…

But I have departed from my initial thought. The thought was that perhaps we must conceive concepts in a different way. Namely, we must give up linking conceptual content (or structure) with thought only. Actions have conceptual content. Their conceptual content is revealed when we give reasons for doing something.

[“Why link conceptual content of actions to the practice of giving reasons for having done something. Why not link it to the practice of providing explanations? You need concepts for explanations as well.” – Right, but you need different concepts.]

In fact, we can think of non-linguistic actions as having conceptual content. When playing Super Hexagon I started to identify patterns in the forms I was trying to avoid. This happened at a non-linguistic level (perhaps because everything was extremely fast). Of course, if one gets used to produce some pattern of reaction to some pattern of stimuli, this does not mean that one has concepts. However, at some point, while reflecting on the game, I was thinking of some specific pattern (I might have used some representation to do so. If so, then I suppose I was using it as a handle for that pattern of forms).


Where am I going with this? I want to say that if the ability to reflect on what you are doing is not necessarily linked with communication, then concepts are not necessarily linked with communication.

“But they must be linked to a language, even if one uses that language to think out loud (as Sellars says).”

I do not know about that. Perhaps the ability to simulate what you did is enough for some primitive form of reflection. The point is that I want to think of pre-linguistic actions as having a conceptual structure. It’s structure can come from pre-linguistic communal practices and so on.

“So your primitive concepts will be reduced to such practices via extra-semantic rules? Isn’t this contradictory?”

Well, the rules are not semantic in the sense that they do not regulate the use of any words. They can regulate the use of arrows in a community which has only a few signals, but not a language. It could be said, nevertheless, the those people have a primitive concept of arrow, even if they can not talk about arrows. What they can do is to observe how arrows are used in their group, to imitate each other, to correct each other, to teach their children to use arrows properly and perhaps other similar things.

The point is not to conceive “thinking” as an activity which can be done only in a language.

“But then you’ll have to say that one ‘thinks’ whenever one performs an action.”

If this seems awkward to you, then you can say that actions involve concepts, since any action has a conceptual content (prior to receiving a description), while thinking is “the use of concepts”.

This could be detailed further, of course. For now, however, I want to return to my initial question. How do we distinguish between presuppositions and logical consequences? I the case of linguistic actions we say that a failure at the level of logical consequences renders what I was saying false, while a failure at the level of presuppositions renders it meaningless. When we move to non-linguistic actions it seems that we cannot distinguish between truth and meaning, since opening a door can only be successful or not. However, there are different ways in which one could fail to do something. One could fail, for instance, in such a way that it would be clear what one was trying to do (score a goal), but the result won’t be achieved (“she tried to score a goal and missed”). In other situations, one could fail in such a way that it will not be clear what one was trying to do anymore (“Was she shooting or passing? But there was nobody there and the goal point was in another direction”).

So perhaps we could distinguish between necessary conditions for an actions (requiring other actions to be performed) such that failure in their case would amount to missing the goal (saying something true) and conditions such that failure in their case would amount to not loosing a goal altogether (saying something meaningful, no matter what).

“Right, but you need to talk about necessary conditions in both cases. The person trying to score a goal and missing was, supposedly, assuming responsibility for doing whatever was necessary in order to shoot, but made a technical mistake. This is not analogous to saying something false, but to mispronouncing a word (or to a spelling mistake).”

A spelling mistake could render what one says false or meaningless. Suppose that if I do not push button B, the light will not turn on when I push button A. However, button A is connected to the electrical circuit and to a light bulb. I can now say that pushing button B follows by reason from pushing A to turn on the light, in the way in which a logical conclusion follows from a premise. And now let us think of a case where I push button A without connecting it to the light bulb (and it is not clear from the setting what was A supposed to be connected to). I would say that connecting A to the light bulb follows by reason from turning on the light by pushing A, but rather in the way in which a presupposition (“John was a smoker”) follows from a statement (“John quit smoking”).

The whole point is to accept that the distinction is not clear cut. Difficult cases which require a practical decision may be encountered.


From a personalist-actional point of view, the question “How can a natural event be someone’s action?” is meaningless. Instead, one has to ask: “How can something be less than an action?”. The answer to this question will perhaps be that there are actions nobody assumes responsibility for and such actions are called natural events. (The way we usually talk suggest that we use the fiction of Nature as agent of natural events.)

There must be an environment acted upon (otherwise, there would be no goals – that is, unachieved states), but the existence of such an environment does only require the existence of some states of things (different from those we are trying to achieve) and not the occurrence of some natural events. The environment is the background upon which my actions can interfere with other actions (performed by myself and other agents), but that is all.

Of course, some agent(s) must be responsible for the existence of the environment, since any result is the responsibility of some person a. s. o. (we do not talk of someone or something having caused the existence of the environment).

The main point I was trying to make earlier is this. Knowledge has to be “our knowledge of the real world”. For this it has to reach to the real world, in a way no though can (conceived in an intellectualist manner / maybe as a psychological thing), but also to be our knowledge, in a way no natural event (lacking an agent and a conceptual content) can.

My answer is that we know things by acting directly in our environment and that our actions have agents (us) and a conceptual content. And now this answer itself could be expanded in either a naturalist or a “personalist” vocabulary. The problem is that when one does that one gets into trouble. It is largely believed nowadays that the naturalist vocabulary is better at avoiding troubles, but I would claim that this view is due to misunderstanding how the personal vocabulary works. For instance, it is considered a problem that one talking about persons cannot clearly say how can a person cause an action, while the personalist would not regard this as a problem. Persons do not cause actions, they are simply responsible for them.

In a similar way chains of means and ends can be “inserted” either in the natural space of causes and effects, or in the logical space of responsibilities (towards actions of using the means, following by reason from the action of achieving the end).

In a sense, the naturalist does not have to justify her vocabulary choice. It is enough to offer a natural explanation for having chosen it. If the problem is that we cannot conceive cognitive processes as “our knowledge of the world”, the coherent naturalist could reply that it is enough to talk like she does for a long enough time and everything will be OK.

(It now seems to me that perhaps one cannot be a “good naturalist” without also being an eliminativist.)

So perhaps the troubles I was alluding to are only the concern of the personalist. She must assume responsibility for choosing her vocabulary and must not do so if she has no reasons for doing it. The reasons, however, could include what this vocabulary enables her to do (if she is a pragmatist, like I suggest she should be).


An example. A study shows that people understand different things by “a happy life” and “a meaningful life”. The traditional philosopher could ignore it. After all she knows that the two are different concepts. She knows it by conceptual analysis. She performed some mental experiments, including, for instance, one in which a person is made happy all the time by being connected to a computer, but while living an absurd life. Of course, for the mental experiment to be successful one must have the intuition that the life discussed was happy and absurd. For this, in turn, one must rely on how those expressions were used within the linguistic community one was a part of, and those usages were precisely what the initial empirical study was about. If the results of that study were different, pointing out, perhaps, that people use “happy life” and “meaningful life” interchangeably, the mental experiment performed by the philosopher would have been a failure, since the intuitions to support its intended outcome (to show the difference) would have been missing altogether.

Now, suppose that a similar study showed that people understand the same thing by “happy life” and “good life”. Perhaps the philosopher won’t be able to show by a mental experiment that the two are different concepts, but she would still be able to propose a distinction between the two, by showing that such a distinction could serve some important purpose. For this, she could imagine a case where a conceptual distinction would be needed, but none was available. For instance, she could imagine a life during which one meets one’s goals without self improvement and convince other people that although this might be a happy life, it might be important to distinguish it from a life during which self-improvement was the condition under which one’s goals would be achieved, and call only the later a “good life”.

This might perhaps require that good was used non-interchangeably with happy in some expressions, but a study showing that this is so would not lead automatically to this conceptual proposal.


Not many are ready to become “good naturalists”. Interest in some use of what I call “the personalist vocabulary” is not lost, although the use in question is rather confused (and confusing). Not only does this vocabulary deserve to be saved, it also deserves to be improved.

“But what can be achieved by that? Even if you clean it from its psychological impurities, you’ll have to connect it to the causal talk (at least), and you cannot do that. You assume that it could be self-contained, but is this really so? You still need to talk about objects and states of affairs and events and causal connections and so on. Is this really a vocabulary, after all? Aren’t you just trying to imitate Berkeley, perhaps to the same result?”

I could say that “object” has a different meaning in the naturalist vocabulary from the one it has in the personalist one. The same could be said about other concepts.

“Right, but what is the world made of, in the end? Is it made of natural objects, having natural properties, or of your artifacts, raw and waste materials, human bodies and so on?”

Well, your question itself could be understood in at least two different ways. If you’re asking me what should science assume as existing when talking about “the real world”, I might talk about natural objects, properties and events (or I might as well decline to answer, since I have no authority in this matter and I am not responsible for telling science what to assume). If you’re asking me what can I assume a responsibility for when talking about the world around, I am going to reply that I can only assume a responsibility for artifacts, objects acted upon, actions and persons.

“You are trying to play a trick on me here. What you consider yourself responsible for is your problem. If you think you have no responsibility to answer my questions, you may reject them, but you do not need to talk about your responsibility when I am asking you something about a completely different topic.”

I do not think the topic was completely different. I was telling you that the world is made of whatever can figure in a network of responsibilities, be they of the scientific community or of the common person.

“A scientist trying to measure the flogiston from an object could be said to assume a responsibility for the existence of flogiston. Yet flogiston does not exist.”

Well, I thought you were asking me about the most general kinds of stuff the world was made of.

“I was. My point was that assuming responsibility for the existence of something does not guarantee existence. This seems to be so not only for the flogiston, but also for the most general kinds of stuff.”

Well, then my answer was perhaps meant for another question: “What could the world be made of?”. Even if we were to “discover” that there are no objects (in whatever sense), we’ll still put in their place (in our ontology) something which can figure in a network of responsibilities.

“Everything seems to be able to figure in such a network. Isn’t you answer vacuous when you phrase it like that?”

I disagree. The answer gets weird when you try to drag me onto the playing field of science.


If minds do exist, the best way in which they could be described is provided by the traditional dualist: they are some sort of objects (or substances) lacking spatial properties, they can only deal directly with mental stuff (perceptions / sensations / thoughts) a.s.o. We are minds and it is incomprehensible that we have bodies and live in some non-mental environment. Also, every mind is completely isolated. A mind can only know (and speak of) its own content.

With the linguistic turn, one could leave aside the connection between a mind and its environment and talk of reference instead. However, the relation of reference turned out to be as incomprehensible.

What basis do I have to believe that some sort of ‘actional turn’ can do better? Well I have started by getting rid of a relation of correspondence between mental representations (or words) and objects. We act directly on objects, even when we speak of them (a name might be an object-handle or a place-holder or an individual tag). Knowledge of our environment also starts with acting in our environment. Sensory input makes it possible, but is no part of it.

However, there is still a gap left between agents / persons and their actions. If we conceive actions as events caused by intentions, which, in their turn, are reducible to neurophysiological events, no matter what specifications we put on the causal chains involved in the production of proper actions (to distinguish them from accidental realizations of intended purposes by failed attempts), there is still one problem left. Namely, such descriptions are perfectly compatible with the lack of agents. Even if we insist that an “intentional event” can only occur in a brain, one can imagine cases in which a brain does not belong to any person. And I do not think of a zombie brain here. It could be a brain in which no transcendental aperception is realized. Different mental events are not integrated into a unity. But even if they were, it might be that the ability to assume zero-level responsibility for the events caused (in the proper way) by your intentions is not developed. Or a brain realizing that ability only for present actions and not past ones (due to some memory defect).

Perhaps from some point on we might speak with ease of actions being events caused in some specific ways by brains functioning in a specific way. At that point, we might be willing to attribute responsibility for an action to a brain in a way such that this won’t mean that we deny our own responsibility for the same action. At that point we would treat brains as persons, perhaps. It seems a bit of a stretch, but I do not want to say that it would be incomprehensible to talk like that.

In any case, the alternative is to deny that a person is something – some sort of object or substance. Talk of persons is in a sense embedded into the way we view our world. The attempt to purify our world of actions performed by persons has produced the concept of a natural phenomenon (or event), but the attempt was misled by its assumption that we purify nature of some obscure entity. Persons are, if you wish, social constructs. They are no more real than any other social roles. Being a person is a basic social role, in the sense that you cannot have other social roles without having this one. You cannot be spoken to, for instance (although you can be talked to), if you’re not a person. You cannot be a parent, a teacher, a student, an employee, etc. without being a person. By saying that something (a body, a robot, a living being) is a person we accept that something into several of our practices. The concept might change, of course. It could separate more and more from the concept of an agent, for instance.

Of course, reducing “being a person” to a social role can also be regarded as some sort of naturalism. Actions, according to this view, are natural events caused (in some specific way) by brains functioning in such a way that their interactions to each other (via bodies and other stuff from their environment) can be described by certain sociological theories. Assuming a responsibility would amount to taking part into certain practices – not avoiding punishment, criticism or praise, for instance (certain theoretical complications are going to be needed in order to distinguish a negative reaction from a punishment, etc.).

Reducing sociological concepts to a vocabulary describing “brains-interactions” might prove difficult, but is not in principle unachievable. So, if one pretends to have the entire picture in front of one’s eyes, one might perhaps say that the gap between us and our environment was bridged. The trick might be to replace the human-specific clear-cut vocabulary with a related one which was not human-specific and admitted of degrees (between being an agent and not being an agent, for instance). This could still seem a bit of a stretch, but could not be rejected on this very ground.

If talk of persons and actions is simply fostered by some particular twirls of some special brain interactions, then everything seems to be solved, from a naturalist point of view. However, the proper description of those twirls might in its turn require us to use “personalist-actional” talk (what distinguishes a punishment from a negative reaction being that the first is directed at a person, for instance). Maybe one could circumvent such problems by using structural / axiomatic definitions for these “personalist-actional” concepts (think of Davidson’s proposal for defining ‘truth in L’, or of Carnap’s attempt to define the Ähnlichkeits-erinnerung relation).


The moral of this story is still going to be: “The world does not really contain agents (or persons), but it contains organisms which are complex enough to engage into interactions which are complex enough to make talk of agents and actions possible. There is no gap between agents and the environment they are acting upon, because there are no agents and actions – there are only natural phenomena and agency is one such phenomena. Now, if one adds the claim that talk of agents, responsibility and actions should not be ‘eliminated’, since it is useful to describe such extremely complex phenomena, we get a view which not very different from Davidson’s anomalous monism (we can call it ‘non-eliminativist naturalism’). My problem with this view, however, is not due to its lack of originality. I do not have a serious problem with its ‘first-person naturalism’ either. Being told that my assuming responsibility for my actions is (in the end) a natural phenomenon does not prevent me from doing it. I have a more serious problem with the ‘third-person naturalism’. This is a twofold matter, actually, since there are two roles involved in the performance of any action: that of an addressee (the one whom I am performing the action for) and that of a “responsibility-holder” (the one who can hold me responsible for my action). Now, the addressee of my action might not actually exist as a person. I might be convinced to accept that the addressee is naturalized as well. The “responsibility-holder”, on the other hand, cannot be reducible to a natural phenomenon, in my view. Perhaps this my incapacity. I know that “being responsible for A” can be described as part of a practice which, in its turn, can be described as some sort of complex “brain-interaction”, but everything seems lost at this level of description. The “responsibility-holder” must be a real person. A super-natural person, in some sense.

This seems too little, of course.



“What is the difference between what you say about responsibility and what Brandom says about commitment and entitlement?”

I would say that opposing actions are such that the performance of one of them prevents one from assuming responsibility for the other (within the same series of actions, of course). In such a case Brandom would talk about entitlement – if you are entitled to say A, then you are no longer entitled to say B.

In the case of B following from A (or being inferred from A) I think Brandom would talk about commitment: if I am committed towards A, then I am also committed towards B. The same goes for entitlement, of course. Inferential relations transmit both commitment and entitlement from a premise to a conclusion. I would say that doing A is a sufficient condition for assuming responsibility for doing B. This does not mean than if one does A, then one ought to do B. In such a case one is simply able to assume responsibility for B (that would be “entitlement”, perhaps). Moving a pawn from a6 to a7 in chess and doing whatever is required to clear a8 are sufficient to assume responsibility for moving the same pawn from a7 to a8 and transform it into whatever piece you choose (apart from a pawn). Brandom would perhaps say that if you are entitled to do the first moves, then you are also entitled to transform the pawn.

On the other hand, while Brandom would say, perhaps, that a commitment towards “All men are mortal” and “Socrates is a man” renders a commitment towards “Socrates is mortal”, I would say that assuming the responsibility for asserting (or agreeing) to “Socrates is mortal” is a necessary condition for asserting both that “All men are mortal” and “Socrates is a man”. So one cannot properly commit to both of them without committing to their conclusion, in Brandom’s words. According to this view, commitment does not go from premises to conclusion, but the other way around. This does not contradict what Brandom says, since he is talking about sufficient conditions and I am talking about necessary conditions. If a commitment towards A is sufficient for a commitment towards B, then a commitment towards B is necessary for a commitment towards A (supposing the equivalence of ‘A is a sufficient condition for B’ and ‘B is a necessary condition for A’).

Also, my talk about sufficient and necessary conditions seems to be incompatible with a transfer of either entitlement or commitment. I might be entitled to move the pawn from a6 to a7 (no rule of the game prevents me from doing so) but this is not sufficient to entitle me to transform it into a queen, by moving it from a7 to a8. What enables me to assume a responsibility for transforming it into a queen is the actual move (assuming that a8 is empty). Since I cannot assume a responsibility for transforming the pawn into a queen unless I can do that.

“There is something awkward here. You sometimes speak as if one could not assume responsibility for A unless one did A, while other times you seem to accept that one could assume responsibility for A without doing A.”

I cannot assume zero-level responsibility for A unless A occurs, but I can assume a duty-responsibility for A even if A does not occur. Of course, I do not want to say that one has a duty to assert whatever follows from any premises, but it could be said that one has a duty to agree to whatever follows from what one asserts.

“Suppose I say that my brother is home and if my brother is home, then the door is unlocked. According to you, it seems that by saying these things I have a duty-responsibility to agree to the assertion that the door is unlocked. What if I get home and on trying to open the door I discover that it is locked? Do I still have a duty-responsibility to agree that the door is unlocked?”

Well, let us say that it follows by reason from my failed attempt to open the door (and my discovery that it is locked) that the door is locked (the belief that the door is locked, so to speak, is supported by my actions. Since my actions now prevent me from either asserting or agreeing that the door is unlocked, they seem to be opposed to my previous assertions that my brother is home and if he is home, then the door is unlocked. So now I have to either open the door without unlocking it first (although my first attempt to do so was unsuccessful), or to withdraw one of my previous assertions.


Let me return to the case of empirical actions, anyway. Suppose that I do this: I grasp the cup on my table and drink some tea from it. I perform this action under this (and other similar descriptions). I would also assume responsibility for grasping the cup, lifting it from the table to my mouth, inclining it, drinking tea from it, etc. I might assume responsibility for reaching towards the cup, but not for moving my hand one centimeter in the cup’s direction. I would agree that this did also happen while I was reaching towards the cup, but I would deny that this was my action. (This should make my understanding of basic actions more clear).

Speaking as we usually do, several beliefs can be supported by my complex action. To list a few:
(i) There is a cup on my table.
(ii) The cup on my table contains tea.
(iii) The cup is not glued to the table.
(iv) I am able to lift the cup on my table.

The performance of my entire action is sufficient for me to assume responsibility for holding those beliefs (whatever this means – asserting (i)-(iv), opining that (i)-(iv), agreeing to (i)-(iv), etc.).

Now let us think of only a part of my entire action – the part where I have grasped the cup and lifted it from the table. Let us call this part the action A and let us call B the action of putting the cup back on the table. Now, I could not put the cup on the table if it was already on the table. In order to put the cup on the table I had to lift it first. So doing A is a necessary condition for assuming responsibility for doing B (so to speak). Doing A, then, follows by reason from doing B. Also, believing that (i), conceived as some sort of action, follows by reason from doing B. But believing that (i) is a necessary condition for assuming responsibility for doing A, so believing that (i) also follows by reason from doing A. I do not see any problems here. (Well, doing A would justify the belief that “there was a cup on the table”, not (i), but I will leave this aside for now.)

Doing B prevents me from drooping the cup on the table (without performing another A-like action). Let us call C the action of dropping the cup on the table. Now, the performance of C would be perhaps sufficient to justify (i). The problem, in this case, is that one could be justified to hold a belief by performing any of two opposing actions. This, in itself, is not unacceptable, since (i) could follow from both [(i) and X] and [(i) and ~X], which are contradictory.

I must think of other possible difficulties with this view.



Basicly, my point seems to be that you can do something in order to justify something else done by you. According to this view, we do not justify beliefs, but assertions (or other speech acts) by performing other speech acts or non-communicative actions. But wouldn’t this mean that we can also justify non-communicative actions by performing other non-communicative actions as well? However, if B follows by reason from A, this does not automatically mean that the performance of B is justified by the performance of A. Or does it?

In addition, perhaps we should distinguish between (i) justifying the performance of action X (under some description) and (ii) providing your reasons for doing X. One’s reasons for asserting p could be different from one’s justification for asserting p (or one’s reasons to believe that p, in other words). For instance, my justification for asserting ‘It is raining outside’ could be that I have just wittnessed the rain outside (by performing certain empirical actions like listening to the sound the raindrops were making, watching them falling etc.) but my reasons for making that assertion to you could be to provide you with useful information on how to dress yourself if you’re going outside. It could be said that one can inform someone that p only if one asserts that p (or at least only if one performs an assertive of some sort). If so, asserting that p to you follows by reason from informing you that p. But of course, the action of informing you that it is raining outside cannot be used as a justification for the assertion that it is raining outside.

And now the problem is this. Since not all the actions X follows by reason from constitute justifications for (the performance of) X, why do I say that an empirical action of the ‘wittnessing that p’ sort justifies the assertion that p? Only because the two have identical (or at least similar content)? We can imagine the following case: a result is obtained by pressing a button three times. The second press could be regarded as a necessary condition for the third press (I will leave aside any details here). Now, even if the second and the third presses do not have precisely the same conceptual content, their conceptual content is similar. However, none of the two presses justifies the other.

Another example: I press the button the third time and say “I assert that I have pressed the button for at least two times”. My pressing of the button might be said to justify my assertion, but the content of the two is different.

[Also, what if I do not remember how many times did I press the button? Even if this was my third press, my assertion was ungrounded from my point of view. A person watching what I was doing could reasonably accept my action under the description ‘a third press of the button’, but if I do that, I do not really have a justification for doing it. So the conceptual content of actions seems to require more than just the performance of some empirical actions by the agent in case. – The reply here would be that nothing was said to reject the empirical action model of justification. The model can be of course extended to account for interpersonal justification, but the most important point is perhaps that pressing the button for the third time is not an empirical action (as opposed to witnessing A pressing the button etc.). – Here the opponent could say that one could wittness something and forget all about it afterwards, so at least “remembering having witnessed A’s performance of X – where X contains the description in case” is an additional requirement, but remembering something is not an action.]



[Intermezzo] What do I really want to say? My main interest is in this special vocabulary one could use to speak about responsibility, persons, their actions (including their use of a language), their reasons, their concepts, intentions, rules and values. However, I do think one could properly speak in such a vocabulary as if there was no nature and causality. Persons can not simply float in the space of their actions, responsibilities and reasons (although I am very tempted to say that from a point of view things stand as if we were floating in a network of responsibilities; this network is what gives meaning to our lives; this includes our responsibilities toward future generations – that is, future persons).

“So, if we knew that the Sun is going to explode in 10 years from today and kill us all, our lives would loose their meaning?”

– Perhaps they will have meaning as long as we hoped to somehow escape from this catastrophe. But even when all hope was lost, perhaps our lives still ought to have some meaning. In a sense, the network or resonsibilities is given. It is as if it did not depend on our existence. The network (conceptually) depends on the possible existence of persons and the possibility of a person’s existence also depends on the possibility of being inserted into such a network. But while using this vocabulary on will talk about the existence of a person in a different way from the one we use to talk about the existence of “natural objects” or the occurence of “natural phenomena”. If needed, another word could be used (“presence”, for instance). One could say, in this way, that a person is present (or “no longer present”). Dead people, we use to say, are “no longer among us”. They “have departed from us” and are “no longer present”. This is entirely different from saying that their brains have stoped functioning. Their position in our network of responsibilities has changed. They will no longer be responsible for performing any future actions. They could still be considered responsible for their past actions (and some of their consequences), but not “held responsible”. Some of our responsibilities towards them have not changed, but no new responsibilites towards them can be properly undertaken.


“I know that…” – Wittgenstein seems to suggest that sometimes we say this in non-epistemic contexts. “I know that he is in the same room with us and can hear me”- does not purport to state any knowledge. By saying this I could acknowledge the fact or express my being aware of it. In other words, saying “I know that…” does not always incur for the speaker the responsibility to justify her “knowledge”.


So, what is the meaning of life? What is it to be alive? What is it to be human? Why is my own life important, worth living? I occupy a unique position in a network of responsibilities. My life is composed of my actions, but those actions have a deep meaning not because they have a purpose, but because I assume responsibility for them.

“But there can no be duty-responsibilities to perform particular actions. The object of a duty-responsibility to perform an action can only be a generic action (“to give your child healhy food, not this particular dinner”). The network of responsibilities could never account for a unique position which only you can hold.”

But only I (this particular person) could be responsible towards my child (this particular child).

“Indeed, but if you were not present, your child would not be present either and this branch of the network would not exist. It is not as if the existence of the network depends on your presence.”

Well, I was not trying to claim that my presence is in some sense necessary. I could have not been present, of course, but the (closest) world (to this one) in which I were not present would be an entirely different world. In that world my presence would be impossible.

“But then you seem to claim that your presence is somehow necessary in this world.”

Well, I say that the presence of this particular person would be impossible in any world except for this one. This would could go on without me being present, but perhaps it could not be _this world_ without me being present right now. In any case, neither necessary existenced, nor uniqueness are needed for my life to have a deep meaning. A network of responsibilities seems to be sufficient.

“Right, but is it necessary?”

Well, I can conceive a world in which nobody held me responsible for anything (perhaps I had the Ring of Gyges or something similar), but that would still be a world in which I would be responsible for my actions and a lot of other things. When I try to conceive myself outside a network of responsibilites, I am not a person anymore and my life becomes absurd.

“So you’re saying that our lives have meaning because we assume responsibilities. But what is their meaning? What is one meant to achieve by living?”

I suppose we are trying to preserve anything valuable left to us by our ancestors and improve the world (make it more valuable) for future persons.

“And how does this, in itself, have any meaning? Why should it be important to do so?”

It shouldn’t be important, perhaps, but it ought to be important, since nothing else can be more important.

“What if nothing is?”

Then that’s it. But it still ought to be important to try to improve the world for future persons. We have this responsibility. I, at least, acknowledge that I am responsible for doing this.

“But isn’t the meaning, then, provided by anything one feels responsible for?”

One cannot restructure the network of responsibilities according to how one feels.

“Even if the world is never destroyed (in any case, I suspect that by “the world” you actually mean “this network of responsibilities), if the process of improving the world goes on and on forever, how can this achieve anything? Isn’t such a process meaningless, after all? And if the process reaches an end, what meaning could the lives of the persons living in a perfect world have?”

Even a perfect world can loose its perfection. Keeping it as such would still be meaningful.

“A world which could loose its perfection wouldn’t really be a perfect world, would it? I mean, a perfect world should require no maintenance.”

You seem to think of a world in which nobody would be responsible for anything anymore. A dream world, so to speak. No person could be present in such a world. Such a world could only contain “non-playing characters”, so to speak. What I am thinking at is something entirely different. It is, if you which, a wold full of value and meaning, which must be preserved and could still gain value and meaning, yet is such that no person would anymore struggle to keep her or his own responsibilities.

“Ok, but this seems to be one utopia among others. Why should the meaning of your life depend on this particular one?”

Well, it doesn’t. My life has meaning because I am a person present in a network of responsibilities. Its meaning is to improve the network in ways I can conceive of. This utopia was one of them.

“Right. Ok, I’ll leave it at that, for now. Still, did you notice that as our talk went on you moved from using a regular vocabulary to using only your prefered vocabulary?”

Since you did notice it, of course I did too. However, I see nothing wrong with this. After all, this is the vocabulary I am trying to develop. It should be useful if one wanted to talk about such matters without stumbling into philosophical problems at every step.


“Would you say, then, that knowledge is a matter of responsibility?”

Yes. Saying “I know that….” in an epistemic context encumbers you with a certain responsibility.

“Would you, then, agree to Eike von Savigny’s idea the truth is a quality of those sentences the assertion of which leaves the speaker (in some sense) unencumbered?”

I still have to think about that idea.