Psychology and responsibility

Some notes written about one year ago. I did revise them, but nothing is publishable.


Wittgenstein’s PLA leads to the conclusion that private psychological states (accessible only by introspection) can not be described or named. This idea leads to the point that one can establish that a person has a responsibility without talking about her inner psychological states. A simple way in which this could be done can be imagined:

(a) X has done A (where A is a description such that X accepts that she/he did A and A was successful).
(b) If (a), then X had the intention to do A.
(c) If X had the intention to do A and X did A, then X is (zero-level-)responsible for having done A.
(d) X could not have had the intention to do A unless X believed that A can be done.
(e) A could be done only if… – here we specify all the necessary conditions for A’s performance: the object(s) on which A was performed, A’s agent, addressee and overseer must exist etc.
(f) X is responsible for believing that all the necessary conditions for A’s performance did obtain at the time at which A was performed (we do not bother ourselves too much with time specifications right now).
(g) Where the necessary conditions for A’s performance include other actions (A is part of a series S) and X did not cooperate with other agents in achieving the end of S, X is responsible for having done all the other actions in S.
(h) X is responsible for believing all the beliefs rationally entailed by X’s beliefs at (f).
(i) If A is part of a series S and X has assumed responsibility to perform S (non-cooperatively), then X is (duty-)responsible for doing the rest of the the actions in the series S. X is also (duty-)responsible for refraining from / avoiding to do any actions opposed (in the series S) to those in S.



What is it for a computer to be intelligent? I think the real problem is obscured by talking about any criteria for intelligence. One can always admit that there could be intelligence without understanding. So, what we ask is actually whether a computer could ever understand something.

We could think of non-human understanding at this point. People could say of a dog, for instance, that it understands several things. This could be said without talking of a dog as if it were a human. We include dogs into some of our practices in such a way that we can effortlessly say that dogs understand things.

We could also think of how we treat a child. We allow the child to take part into our practices gradually. Some practices assume that all the participants understand some things. Now, we could re-describe some practices in such a way that they wouldn’t require understanding (“Playing chess only requires a mechanical application of the rules of chess”) or modify our vocabulary in some other way (“A computer does not play go, it only performs computations which can be translated into go moves” etc.). At some point such re-descriptions might require too much effort from our part, so we might give up such attempts and start talking about computers taking part in those practices as “understanding”, “knowing that”, “intending to” (see how Hassabis talks about AlphaGo).

So, could we know in advance where that point is? My answer is negative. There are differences between cultures. Some cultures might accept artificial agents into some practices more easily than others. Some might accept them, but assign a different social role to them (not that of agents, for instance).

A few suggestions could still be made. One is that we might need to figure out how to attribute responsibility to an artificial agent. This does not seem directly related to understanding, but in my view it is (if one talks about meaning in the framework of an inferentialist semantics and replaces talk of commitments and entitlements by talk of responsibilities this becomes obvious).

(Think, for instance, of the following scenario. When AlphaGo makes mistakes and looses in front of a strong player, we only let it play with weaker players for a while. This is a consequence of making mistakes. AlphaGo can evaluate the strength of its opponents and conclude that they are weaker. It can also relate this result – facing weaker players – to what produced it – making crucial mistakes in a game against a stronger player. After a while it can “develop a sense of responsibility”. This is a crude scenario, of course.)


“If you opine that p…”

I would perhaps agree that believing is not opining. Opining is the performance of a particular speech act. Believing, when it is not an alleged psychological state (a propositional attitude a.s.o.) amounts to a duty-responsibility to accept some belief-attributions. This looks circular, of course. What is a belief-attribution? How can one accept or reject a belief attribution? Couldn’t one agree to ‘So you believe that p’ but reject p? – Perhaps one could do so, on pain of being accused of irrationality. One could, for instance accept that she has an irrational superstition. In such a case, one could opine that ~p and accept a p-attribution of belief.

Still, what is a belief-attribution and how could one accept it. We can usually accept or reject invitations, but a belief-attribution is not similar to an invitation. It is also different from the question ‘Do you believe that p?’, which is the request to either state that you believe that p or you don’t (of course, one could reject the request on various grounds, and leave the question unanswered). So perhaps a belief attribution is a statement of this kind: ‘X believes that p’. If I am X and agree to this statement, then it can be said that I accept the belief-attribution. If I disagree with it (or deny it), then it can be said that I reject the belief-attribution. But what does the sentence say?

The sentence could (in principle) say (i) that the state of believing that p occurs in my brain at present, or (ii) that I have the disposition to have that brain state under the appropriate circumstances.

It is obvious that a belief-attribution does not say (i). It could say (ii), were it not for unconscious beliefs. That is, when one attributes a belief to X one does not exclude that case in which X would not be aware of believing that p and (ii) would not obtain. X might not be inclined to say that p or agree to p, but one could still attribute p to X based on other actions performed by X. So we are back to ‘X believes that p’. The sentence seems to mean that X is duty-responsible to opine that p in the appropriate circumstances. So in the case of an irrational superstition one is aware of the clash seems to be between the duty to opine that p (in the appropriate circumstances) and the act of opining that ~p, which is opposed to that of opining that p (in normal cases) and thus renders one unable to fulfill one’s duty-responsibility.

In short, opining that p would seem to mean that your are zero-level responsible for opining that p, while believing that p would mean that you are duty-responsible for opining that p.

No ontological commitment is needed in order to talk about responsibilities. Responsibilities cannot be psychological states, since we would have an infinite regress if they were (one must have the responsibility to have a responsibility etc.). Also, they cannot be caused, but one can only have them. We can also talk about someone assuming a responsibility, but this can be discarded at this point.

What does it mean, then, to have a responsibility? Doesn’t it mean that one can be held responsible for whatever one is responsible for? The etymology suggests that being held responsible is being in a position such that you must answer to something. What is one supposed to answer to? Some requests (or requirements), of course. This keeps us in a vicious circle as well, since “must answer” means “has a duty-responsibility to answer”.

We are looking for a realistic model for rules. The model would still be a model, though, and not a description of some facts. Rules seem to correlate duty-responsibilities to zero-level responsibilities (given a network of social roles, which can also be defined by the duty-responsibilities they incur and the actions required for assuming them). We seem to ignore the fact that social roles also are packed with some entitlements (which are not responsibilities). The entitlements, however, do not seem important. At most, one could say that what really matters is that the actions required for assuming a certain social role might be necessary in order to assume some zero-level responsibilities.

So we try to reduce rules to responsibilities, in short. However, if, when pressed to say what is it to have a responsibility, we answer by invoking rules (“must answer” etc.), nothing is gained.

Let me try this. I say that I am responsible for writing these sentences here. What does it mean? It seems to mean at least this. If there was a person entitled to ask me who wrote these sentences and I had no reason to deny that person her request to answer, I would say that I wrote them. However, this not enough. I could forget about writing these sentences and say that I do not know who wrote them, but I would still be responsible for writing them. This means, of course, that I have a duty-responsibility to answer that I wrote them.

This seems to work as a way of reducing zero-level responsibilities to duty responsibilities. There is only one thing left: What is it to have a duty-responsibility? This seems to be precisely what is captured by the phrase “you ought to”. Now, I do not have to acknowledge that I ought to do something in order for it to be the case that I ought to do that thing. Still, it is not a fact that I ought to do A (under such an such circumstances).

“X ought to do A” has nothing to do with predicting X’s action.

Couldn’t we define this “duty-responsibility” axiomatically? That is, couldn’t we build our model by using this as a primitive element in our vocabulary and then say that the only thing satisfying our model is duty-responsibility? Well, even if we could, I don’t feel satisfied by this approach.

Let my try something else. I promise to you that I will visit you tomorrow. I now have a duty-responsibility to visit you tomorrow. What can I be said to have, in other words? The tendency to visit you tomorrow? This is not crucial. The tendency to feel remorse if I do not do it? Maybe so, but this is not crucial either (it could be rather said that I have a duty responsibility to criticize myself and apologize to you if I do not do it, but this is not helpful). It is not that I have something. The world in which I visit you tomorrow is better than the world in which I do not. It has more value, so to speak. This is still not helpful if we want to say what a duty-responsibility is, but it might help us see that there is no ontological commitment in what we say when we talk about duty-responsibilities.

By making a promise to you I exclude some possible worlds from the set of the future worlds I want to live in. However, this is not about a wish. What is a network of responsibilities, then? It would be simple if one could render some future states impossible by making a promise. This is the Wittgensteinean intuition that rules are conceived as super-causalities. One should cease to exist if one attempted to utter a contradiction (Nozick seems to say something along this line, too).

A line in the middle of the road. From a certain point of view, it appears that two cars (coming from opposite directions) are going head on towards each other. Only the line (which is the expression of a rule) prevents them from crashing into each other. None of the two drivers fear that something like that could happen, as if the line was an indestructible wall between the two cars.

In any case, this is not helpful if we conceive such alethic (or metaphysical) modalities (possible, necessary, contingent etc.) with the aid of deontic modalities.

I undertake to visit you tomorrow. This is not something I actually do (an action). What I do is to promise something. By making that promise I undertake to visit you tomorrow, but this is not a different action. Could we say, then, that having a duty-responsibility is acting as if you made a promise? We could, of course, but is this helpful?

Well, perhaps it is, since we can list the verbal actions which we call “a promise” without talking about any duty-responsibilities. Assuming a duty-responsibility, in this case, would always be to make a promise or to act as if you have made one. Having a duty-responsibility would mean to be in the position of a person who has made the promise.

It could now be said, for instance, that when you perform the action A, it is as if you made the promise to answer someone entitled to ask you who did A by indicating yourself. This is the zero-level responsibility. Nothing else, apart from A, needs to be done by you.

“What if this is the way in which I regard an event E, which was not cased by anything happening within my body? Should E be called my action?” – Well, perhaps we do not invoke our knowledge of causal laws when we attribute actions to ourselves and others. What matters are only our beliefs with respect to what cannot be under our control. You could believe that “flipping a coin and getting tails” was something you did. From your point of view, that could be a description of your action. I am going to point out that if by “flipping a coin” you mean “randomly flipping a coin”, then “getting tails” cannot be a part of your action.

“What about intentions? If I buy myself a pair of shoes with the intention to refrain myself from donating money to a particular charity, couldn’t I say that I am making some children in Africa starve to death?” – You could “make them starve” only if you prevented them from eating. The event you are talking about (your spending money on shoes and not on charity) is causally connected to the fact that children in Africa will receive a little bit less food from donations. The amount is insignificant, though. Your intention could not be to make children in Africa starve to death in such a case. I am not talking of your intention as a psychological state, of course. However, this case is pretty difficult. I am still trying to see how I could use my proposed vocabulary to talk about simple cases.

“What is the point, though?”

The point is to avoid some philosophical troubles. Also, to come with a conceptual proposal which makes it easier to talk about us as a part of nature.

“You are making our ability to promise essential to our having rules and social roles, in a way similar to that in which Searle makes our ability to issue declarations essential to our having social institutions.”

I am not sure about this. I do not want to say that our ability to make promises came before our social roles. They might have grown together from our ability to recognize and predict patterns of behaviour.

“But then making a promise is more like encouraging a prediction (or guaranteeing its result).”

I don’t know what to say about this. The point is that once you have the naturalist picture you can forget about it and focus on responsibilities. At that point, our network of responsibilities articulates our “logical space” – that is, the space of the relations of the “following by reason”-kind between our actions (no matter their type) and not a “pure logical space” (characterized by logical implications between sentences).


Basically, what I try to say here is that the language in which we justify that we know something or that some action is reasonable does not need to include any psychological terms.


“X happened due to me, but I did not intend to do X” – So, did you do X or not?

Case 1: You left your phone on the floor, I accidentally stepped on it and broke its screen. I did not intend to break your phone, that is clear. I can assume a responsibility for the fact that your phone is now broken. That is, I am going to assume a duty responsibility to repair your phone. However, I could deny that I have a zero level responsibility for breaking your phone. I did not perform that action. The only action I did perform was to walk across the room. I did not know that your phone was on the floor, so I could not intend to break it by walking across the room (here one can see that my talk about “my intentions” is not a talk about psychological states).

Case 2: I knew that you use to leave your phone on the floor, but I did not look to see what I was stepping on. I could have taken precautions to avoid breaking your phone (this is true for Case 1 as well). If I intended to avoid breaking your phone, I ought to have taken those precautions, which I did not. So perhaps it could be said that I did not intend to avoid breaking your phone. Does this imply that I intended to break it? Suppose one throws a stone at you, while you are not paying attention to what is happening. I am standing next to you and it’s quite easy for me to catch the stone (or at least to divert its path). Nevertheless, I do not prevent the stone from hitting you. Perhaps I act like this because I hate you, but my motives are not important right now. I am responsible for the fact that you’ve been hit by the stone, but could it be said that I intended the stone to hit you?

Well, one might speak like that. If I claim that the object of my intentions can only be an action of mine, I am going to deny this. If I can be said to intend that any event under my control should occur, then I intended that the stone should hit you. So in Case 2, it could be said perhaps that I intended that your phone should break. However, I could still deny that I am zero-level responsible for breaking your phone. In other words, I could deny that I did perform the action of breaking your phone (or the action of stepping on your phone).

Case 3: You leave your cup on the corner of my desk, where I usually put my cup of coffee. In order to put my cup in its regular place I have to push your cup aside. In doing so I do not care if your cup falls from my desk or not. Your cup falls from my desk and gets broken. Using the previous reasoning it could perhaps be said that I did intend your cup to fall from my desk, but now it could also be said that I pushed your cup from my desk and perhaps even that I broke your cup. If so, I am zero-level responsible for breaking your cup. But did I intend to break your cup?

How could I have broken it without intending to do so? Perhaps it was a mistake to use the previous reasoning in this case. After all, not caring weather your cup falls from my desk or not means that I did not know what will happen to your cup. I intended that the stone should hit you in Case 2 precisely because I knew that the stone will hit you if I do nothing and the event was under my control. Here the event is under my control (I could have prevented it from happening), but I had no knowledge about what will happen.

Case 3b: I know that you cup will fall from my desk if I push it aside with my cup, in order to place my cup where it usually stands. I do not care that your cup will fall from my desk, so I put my cup in its place and in doing so I push your cup from my desk. There is no point to say in this case that I intended the event to happen – your cup falling from my desk, since it can be directly said that I intended to make your cup fall from my desk. I am zero-level responsible for making your cup fall from my desk as well and perhaps I am also zero-level responsible for breaking your cup.

“It seems to me that you want to disregard the psychological intentions and talk about intention attributions which coincide, after all, with attributions of knowledge (or beliefs).”

Let me see. I would say that there are no actions, but only action attributions. Zero-level responsibility is not an entity. There are only zero-level responsibility attributions (or assumptions, but these are not acts). So, let us think of the following generic case:

a) The event E is occurring.
b) I believe that the event E is under my control (I am duty-responsible to opine that E is under my control – i.e. that I could either be zero-level responsible for E or for preventing E from happening).
c) Either I do not prevent E from happening, in which case I intend E to happen, or I am zero-level responsible for E, in which case I intend to do E.

Now, ‘I intend to do E’ comes down to attributing the intention to do E to myself. If we remember what we have talked about beliefs, we can now say that ‘I intend to do E’ can be analysed as a duty responsibility to attribute the intention to do E to yourself.

So ‘I ought to say that it was my intention to do E’ can now be broken down to the following duty responsibilities:
i) I ought to opine that E was under my control – that is, to opine that I could indicate myself in reply either to the question ‘who could have been responsible for E’, or to the question ‘who could have been responsible for preventing E’ (where ‘preventing’ means ‘being responsible for an action opposed to E’)
ii) I ought to indicate myself as an answer to the question ‘who is responsible for E’ asked by someone else

“You are trying to analyse ‘producing’ and ‘preventing’ in terms of zero-level responsibility. Why do you avoid to talk about causal relations? After all, ‘E was under my control’ intuitively means ‘I could have caused E or caused something such that ~E’. Your zero-level responsibility seems to be devised only to replace talk of causal relations. Also, if you analyse it in terms of duty responsibilities, why say that E was ‘under your control’ and not ‘your responsibility’?”

My (duty) responsibility with respect to an event which I believe to be under my control is not related to the occurrence of the event in cause, but to an exchange about the event. If somebody asked “But who could have stopped the stone?”, my duty responsibility would have been to indicate myself.

“This is not what you say at (i). There is a duty responsibility to opine that the event was under your control which stems from believing that the event was under your control. The content of that belief, in its turn, includes talk about a duty responsibility to assume zero level responsibility with respect to E. So, if somebody asked ‘But who could have stopped the stone?’, your answer, according to (i), ought to be (due to your belief): ‘If one asked who was responsible for the fact that the stone was actually stopped it would have been possible that I ought (due to zero level responsibility) to indicate myself in reply’. This awkward reply appears only because you want to avoid talking about causal relations. The initial question (“Who could have stopped the stone?” is expressed in causal terms, after all.”

Well, I think here is precisely where you err. The initial question is not expressed in causal terms. I hear the question saying ‘Who could have been responsible for the stopping of the stone?’, so there is no need to answer that “if the question was asked I ought to indicate myself in replay”. The question was already asked so I can indicate myself in replay.

“Ok, but a question can be expressed in explicit causal terms. If I ask ‘Could you have caused something to stop the stone?’ you are forced to give me the long reply, according to (i).”

Indeed I am. And now I think we can see more clearly what is at stake. You claim that when we talk (or exchange beliefs) about what was in our control and what wasn’t we look directly at causal relations. I say that we look at what we could have been zero-level responsible for. Instead of a crucial experiment, we can use the following case:

Suppose I suffer from epilepsy and I know it. I am held prisoner together with George and a sadistic guardian forcefully connects George to a mechanism which is going to kill him in 10 minutes, but can be stopped if an epileptic seizure is detected. Now, if you ask me who could have caused the mechanism to stop from killing George, I can perhaps answer by indicating myself. However, since I cannot go into an epileptic seizure whenever I like, I cannot indicate myself when you ask me who could have performed the action of stopping the mechanism.

In short, I claim that when we talk about events under our control we mean events which we could control by our actions, not events which we could cause to happen (or cause not to happen).

“Right, but now you need the concept of ‘action’ in order to say what is it for an event to be under your control. Also, apart from the circularity which seems to ensue from this, you still need to talk about your actions causing (or preventing) events. All is lost and nothing is gained.”

Let me imagine another case for you. Suppose we have the following events:

(E1) A hand (H1) presses button B1.
(E2) Switch S turns on, letting electric current to flow through wires W1 and W2.
(E3) Electricity in W1 and W2 powers a robotic hand (H2) which moves to press button B2.
(E4) An electric bulb is lit.
(E5) Another hand (H3) presses button B3.

Suppose, in addition, that E1 causes E2 to occur, E2 causes E3 to occur and E5 causes E4 to occur.

(E1)-(E5) are events in a “physical space”, so to speak. If I know perfectly well their causal relations, I cannot assume any responsibility with respect to them. You might think that more knowledge about how I am related to those events (H1 is attached to my body, H2 and H3 are not etc.) will enable me to establish if I am responsible for some of those events, but of course this will not do. More knowledge can be added: I did actually cause the move of H1. Am I responsible for E1? I do not know. Even more knowledge is added: (E1) was caused by (E0) Some neurophysiological processes which could be called “having the intention to move H1″ (as a psychological state) went on in my brain. Am I responsible for E1? I do not know. Can we add more? Sure, (E0) was caused by myself.

This means that I caused something in my brain to happen which, in turn, caused the movement of H1 to press B1. Am I responsible for (E1)? How should I know? After all, if I was not responsible for causing the occurrence of those processes in my brain (or of the psychological state of intending to move H1)…

But now suppose that I assume (zero-level) responsibility for (E1). Sure, if I do, I might believe that I have caused (E1), but that is not important. I might not know that (E4) was caused by (E5), believe that it was caused by (E3) and also assume responsibility for (E4). If somebody asked ‘Who has turned this electric bulb on?”, I would answer by indicating myself. Now, you could show me that I did not cause that to happen or you could assume responsibility for (E5) and know that (E5) did cause (E4) and assume responsibility for (E4) and we might negotiate our responsibility claims to the effect that I drop mine and you keep yours without talking about causal relations. If our responsibility claims were validated by repeated actions “If you say you did it, then do it one more time, etc.”, perhaps everything would be ok.

The point I am trying to make is that there can be no responsibility without placing (E1)-(E5) in this other “space”, which is not the “physical space”. I have called this other space a “network of responsibilities” and have also claimed that a “space of reasons” can be best conceived as “growing” within such a network.

If (E1)-(E5) are placed in this network of responsibilities, I could say: “I can do (E3) by doing (E1)” and assume responsibility for (E3). You can challenge me by saying: “Ok, do (E3) again”. If I did it again and again, you might still believe that it was a coincidence, but your challenge was met. If there is no other responsibility claim for (E3) apart from mine and I can repeat what I claim to be my action, there is no need to be sceptical about it. There could also be attributions of responsibility which I might want to reject by saying that I did not believe that (X) was under my control, so I could not have done (X). Since believing comes down to a duty responsibility to opine that…, you might reject my claims by saying that given what I have previously done (and said), I had a duty responsibility to opine that (X) was under my control, so I did believe that (X) was under my control. You might, for instance, point out that I have interfered successfully with (X)-like events in similar situations before. If I reply that I have thought all those cases to be coincidences, you will reply that an unrealistic dose of skepticism is assumed by such an attitude (in view of the first part of this paragraph) and so on.

The network of responsibilities can produce a way of conceiving events in terms of causal relations. This is what I am trying to say.

“Even if everything you said was right, your network could only enable someone to conceive actions as being causally connected. That is, only those events which we conceive as possible actions of ours could be accounted for by your network. A person talking about events in terms of responsibilities could never get to say that some chemical processes occurring in a person’s retina cause electrical impulses in the connected nerve endings.”

Well, nobody could say such things without performing experiments during which one would describe what one was doing by saying things like “Now I make this chemical process occur in the retina”, “I now trigger an electrical impulse in the nerve endings”, etc.

You seem to view things as if there were natural events and causal relations first and I was just trying to make them fit into a network of responsibilities afterwards. My talk of a network of responsibilities is similar to Heidegger’s talk of Sorge. If one wanted, one could equate a basic form of duty responsibility with Sorge. One could call this an attitude. Our world is met with this attitude – “what am I duty responsible for, in relation to this?”. There is no world outside this attitude.

“What about primitive human beings which do not have the concept (or the social institution) of making promises? You did say at some point that having a duty responsibility is ‘as if one made a promise’. Such primitive human beings cannot have duty responsibilities. Aren’t they also in the world? Isn’t their world our world as well?”

Well, I can, of course try to imagine a world in which living beings from the homo genus cannot make promises. It is not a world in which I (who come from a society which has promises) would not think of duty responsibilities, but I can somehow try to ignore my position in that world for a moment. If I do so, then such a world appears to me as something from a dream. It’s inhabitants are not persons and in a sense cannot be said to exist. The same is true about the entire thing. It is not a world, but only a dream. The “world” of being unable to perform actions is but a dream to me. My cat does not live in “the natural world”, but in a dream, so to speak. So there is no point in speaking about “my cat’s world” and asking myself whether such a world isn’t also my world.


So now let me return to our initial problem:
“X happened due to me, but I did not intend to do X” – So, did you do X or not?

Suppose that I say, with respect to the epileptic seizure scenario mentioned above, that “George’s killing was stopped by me and I did intend it to stop (I had the psychological state of intending for it to stop)”. I could still say that I did not stop George’s killing. It stopped due to me and I had the intention that it should stop. Even if my psychological state was the one causing my epileptic seizure, which in turn made the mechanism to stop, it would be wrong, from my point of view, to assume zero-level responsibility for having stopped George’s killing. If I could learn to induce myself into an epileptic seizure by getting into the psychological state of intending to save somebody, maybe I would say, from some point on, that I can stop a similar killing, but in the first cases it wouldn’t be like that.

If “I did intend to do X” is not meant to say something about a psychological state, but is an intention attribution, then an intention denial might be enough to say that I did not do X, regardless of the causality of X. It could also be noticed that we initially tend to regard the events with respect to which we can attribute an intention to ourselves as our actions and only afterwards correct this tendency.

This reminds me of a funny episode. My son was still small. He was trying to put out a candle (at his birthday) by blowing on it. I suppose his belief was that he can do it (that extinguishing the candle was under his control) and the intention to blow out the candle could have been attributed to him (even by himself, on simpler grounds than those presented here, perhaps – “I am trying to do it, so of course I intend to do it”). I was also blowing on the candle and put it off myself, since he did not seem able to do it. He immediately said: “I blew it off myself!”, which made all of us burst into laughter. I now think he was sincerely issuing a responsibility claim for blowing off the candle. One could say that the problem was his lack of understanding the causality of the entire affair. I tend to think that one gets a better understanding of causal relations via negotiations of responsibility claims. The situation did not necessary require an explanation of why a stronger air flow (as that produced by myself) could rather cause the flame to go off than a weaker air flow (as the one produced by my son). It did, however, require another responsibility claim – “It was me the one who did it and not you”.

“Still, you talk as if ‘I have caused X to happen’ and ‘I made X happen’ where worlds apart, when in actual fact they are used as if they were pretty close.”

People can talk as they want, of course. What I try to do is different. I want to emphasize the difference between causal talk and action talk. There are several semantic rules which can emerge from this difference. What matters is that the rules should fit our intuitions and prove useful, not that the phrases which I use in order to make the distinction apparent should fit our common usage.


It’s not that I feel responsible for my actions and that is what makes me a person. I am a person and I ought to be responsible for my actions. However, if I try to pin this down on some particular experience (the experience of feeling responsible for assuming responsibilities for my actions, for example), it immediately seems to me that the experience in case could be produced in me in some way such that it would be disconnected from my obligations (or responsibilities). I could feel that I have a lot of responsibilities and still assume none. Supposedly, a mad human could experience a strong obligation to assume responsibilities for whatever happened around him and yet never act on that feeling for several insane reasons (or distort the meaning of “assuming a responsibility” beyond recognition; or simply be irrational in his behaviour, no matter what he felt).

No experience can cause a responsibility (or the duty to assume a responsibility) without fault. In a dream one can have all sorts of experiences and yet be under no obligation to assume any responsibility. Not only is the experience of a dream the experience of what is it like not to be a person, but there is no experience of what is it like to be a person (and cannot be one).

Personhood is a social status, one might say. So how do we grant it to somebody? We cannot grant it based on feelings or experiences. We grant it to those who accept both praise and criticism, are able to provide reasons for their actions, to apologize for some of their actions, to promise to do or refrain from doing certain things, try to make amends, improve themselves and so on. This is not to mention some patterns of behavior. This is already a list of actions. We assume that other humans do in fact apologize when they display some “apologizing behavior”, but in view of other actions we attribute to them we can drop that assumption. However, the entire process depends on granting other humans the role of action performing agents. This we do without any need for a proof. We are ready to do it with animals too and we were also ready to view natural events as the actions of some hidden agents some time ago.

It is very strange indeed that we were able to doubt that we are responsible for our actions (see the free will problem). This seems to be an outrageous misunderstanding and perhaps it should be explained. However, I am not able to think of an explanation for this right now.



1) What is an action? (an event one assumes responsibility for as an agent, having an addressee in view, facing an overseer etc.)
2) How can there be logical relations between actions? (negation, conditional actions, material inferences, general actions etc.)
3) Non-communicative actions and communicative actions – How are they related? (through responsibility transmissions)
4) Language meshes with our life – How exactly? (inferentialist-style, but inferences are replaced by ‘following by reason’ relations between actions)
5) Does this conceptual proposal solve any problems? (it might provide solutions to problems like: What is it to be a person? What is the meaning of life? etc.)


“Only you can know if you had that intention” (Wittgenstein, PU, 247) – I push a button, under the impression that it will turn on the light. Instead, some music starts playing. “So you’d like to listen to some music”, you say, but I do not want to be zero-level responsible for turning on the music (I reject that description for my action). I might say: “That was not my intention”. Now, if Wittgenstein is right, this is not the result of some sort of introspection. It is not like I have looked deep into myself and noticed that the intention in case was missing when I was pushing the button. I do seem to have, however, some sort of special authority here. I am the only one who can provide the conceptual structure of my action (it does not matter if I do it or not, or when I do it). Is this epistemic authority? (Anscombe seems to believe so) This could be disputed.

So how do I “introduce” some conceptual structure into my actions, if this is not a psychological act itself? Pouring water into a glass has this conceptual structure (that of being the action of pouring, which can only have a liquid as the object used and some spatio-temporal particular as the object acted upon, etc.) due to our practices (which include what we say not only about actions, but also about similar events – “the water was pouring from the paper cup onto the ground, etc.”). There is no need to intend to pour water into a glass (as some psychological act), but if I try to do it and fail (unknown to me, the bottle was empty etc.), I can say what action I was trying to perform (see Anscombe about the object of an intention being only an action of mine – I can wish for something to happen, but I can only intend to do things myself). Now, it might seem that I am describing an intention I had at the moment of my attempt – the intention to pour water into that glass, as if I was saying to myself (privately) “now I am going to pour some water into this glass”, etc. (this could also be regarded as an intellectualist fiction).

Practices change over time. Think about what it means to write something today. I could write an entire sentence only by selecting words shown to me by a typing software on my phone. Sometimes I chose the next word because it is easier, although I am reluctant to use it. I do not have an alternate word “in mind”, but it does not feel very important and I don’t have time to type a better chosen word. Everyday written communication can be done like this. I do not think I could write notes such as these right now in the same way, but if somebody did do it, I would still be ok with saying that that person wrote down her thoughts.

“But if the conceptual content is established by common practices, the individual does not seem to matter. So how did we get to talk about ‘private intentions’?”

I suppose if everybody was only having correct beliefs…

But we also have other uses for “intention”. When planing a course of action one might present it with the words “this is my intention”. Also, sometimes “this was not my intention” might just mean “I did not plan for this to happen like this”. Only I can know what I was planning to do, but this is not due to some special access, by introspection, to my “intentions”. A plan can be considered a simulated action. The fact that it is a mental simulation is not important here. Suppose I plan to write a letter by writing down its structure, some expressions to be used in it and so on. My plan could be recognizable, especially by someone who knew me well. However, I am the only one who could abandon the plan and describe what I was doing differently – “I was exercising writing a letter, but I did not intend to write one”.

“So what are you trying to say, that your privileged epistemic access to your own intentions is actually only your privileged agency with respect to your actions?”

I am not sure whether to put it like this. There is a link between zero-level responsibility and intentions, so to speak, so neither has anything to do with psychological states. As for “Only you can know if you had that intention”, this does sound for me similar to “only you can be zero-level responsible for your actions” – a grammatical sentence, in Wittgenstein’s words. Different people could assume responsibility for the same event, but only I could assume responsibility for _my actions_.

“Somebody else could provide reasons for your actions, though.”

Indeed, and I might even be convinced to accept those reasons as reasons for my actions (some might be related to some beliefs I was holding without being aware at the moment, etc.). In ordinary cases of non-cooperative actions, however, the other person does not have a duty-responsibility to provide reasons for my actions. And if the action was _mine alone_,… I mean, somebody might be paid to provide reasons for my actions. In this scenario, that other person has an obligation to provide reasons for whatever I do. Even in such a case, only I have the power to cancel the other’s responsibility to say what reasons did I have to turn on the music, for instance, by saying that my action was actually a failed attempt to turn on the lights. In a similar way, I might perhaps cancel your responsibility to say why did I do X by undoing it.


“What about this case? (see Aristotle’s Nichomachean Ethics, 1135a 25-30) You strike a man which you recognize as a man, but you do not recognize him as your father. If you are not responsible for striking your father, this has to be due to how you recognized the person in front of you and this process of recognition is psychological. So how can you say that nothing psychological is involved in establishing your responsibility?”

Well, I had some sensory input. This I think can be described in non-psychological terms. My sensory organs were stimulated, some brain activity happened as a result and so on. This is, however, irrelevant for establishing what action I am responsible for. My beliefs with respect to my environment are relevant, but they need not be conceived as psychological states. In their turn, these beliefs are based on empirical actions. I did not strike my father because I did not believe that the man in front of me was my father. This is because I did not watch my father. That is, the conceptual content of my empirical action was not that of watching my father.

Now, perhaps, you might want to distinguish between:

(a) I was looking at my father (my head was turned towards the man which can be correctly described as “my father”, my eyes were open, I was aware etc.)
(b) I was watching my father (I did perform the empirical action of watching under the description “watching my father” which I would accept as a description of my action)

From my point of view, (a) is not an action. It only says what was happening, not what was I doing. So, I can only say that (a) was happening, but I did not do (b). This is unusual, since usually when (a) is happening I also do (b), but that is all.


”How would you distinguish, then, between intentional and voluntary actions?”

Following Aristotle, I suppose one might say that A’s ”voluntary actions” are events caused by A, which are under A’s control. Aristotle seems to add that A must know what she is doing (so she must accept the regular description of the event in case), but this seems unrelated to the concept of a voluntary action. For instance, doodling could be considered voluntary, even if the agent does not pay attention to what she is doing and does not “know the particulars” of what she is doing. (I think Anscombe would say that doodling is usually voluntary and unintentional).

I still have to think of my answer, though.


“Why do you have to complicate things so much? If you want to say that being responsible for something is not a psychological state one can find oneself in, that is simple. Why not just say that you are an externalist with respect to moral obligations?”

First of all, my concept of responsibility is larger than what that of a “moral obligation”. Second, I do not want to say that being responsible for an action is a state at all.

It is the same with being friends…

“But then you actually want to say that responsibility is socially constructed, don’t you?”

The concept of responsibility is socially constructed, I agree. That means, the uses of phrases like “I am responsible” a.s.o. have been established through human practices, which, in their turn, included various social interactions etc.

I cannot focus very well on these matters.


In a way, my aversion towards psychological explanations for responsibility, values, meaning a.s.o. is not new. I do not want to talk about responsibility or value facts. If my responsibility towards my actions is somehow reduced to some social facts, then so be it. I do not know why I am not bothered by such a reduction. Perhaps because social facts are not facts at all. We use a lot of metaphors to talk about such “facts” („entering” and „leaving” a friendship or some other relation etc.), but we know they are only metaphors.