1 Introduction

A widely shared assumption in the literature on first-person thought is that the competent use of the first-person pronoun and the ability to entertain first-person thoughts expressed through the use of it require prior non-linguistic or non-conceptual forms of self-consciousness, i.e. forms which do not depend on the exercise of any linguistic or conceptual abilities. Many philosophers – such as Bermúdez (1998), Guillot (2016), Lang (2010, 2019), Musholt (2015, 2019), Nozick (1981), O’Brien (2007), Recanati (2007, 2012) and Zahavi (1999) – maintain that the distinctive awareness which accompanies the use of the first person already presupposes a consciousness of the fact that oneself is the owner of a first-person thought or that oneself is the user of a first-person utterance. I call this argument The Argument for Non-Conceptual Self-Consciousness Based on the Meaning of “I” (henceforth also “the argument based on the meaning of ‘I’”).Footnote 1

The central aim of this article is to discuss the argument based on the meaning of ‘I’. After having exposed the structure of the argument (in Section 2), I will demonstrate (in Section 3) that most proponents of the presented argument fail to establish their conclusion, even though I believe that the conclusion is actually true. The argument only justifies the claim that contextual information is needed in order to recognize oneself as the referent of an occurrence of the first person in thought or speech. However, it does not justify the thesis that this information needs to be given consciously and thus the existence of any non-conceptual self-consciousness. Put differently, the argument may justify the claim that first-person thought requires prior non-conceptual self-representations, but to justify that these have to be conscious one has to find evidence on independent grounds.

In the second part of this article (Sections 4 and 5), I shall present an argument for the assumption that some form of self-consciousness is probably needed to reflexively think about oneself via the first-person concept and that subpersonal processes alone might not be able to fulfill this function. In order to do so, I will draw on the general discussion about the functional role of consciousness. In particular, I will rely on results recently proposed by Frith and Metzinger (2016), Frith (2010, 2011), Graziano (2013, 2017), and Graziano and Kastner (2011), which suggest that one of the central functions of consciousness is to provide information about mental states for communication rather than being essential for complex problem solving or decision making.

2 The Argument Based on the Meaning of ‘I’

It is pretty complicated – if not impossible – to show that something does not exist. Nevertheless, very often it seems even necessary for scientific progress to do so, at least in those cases where we want to point out a certain research gap. In what follows, I will attempt to demonstrate that the conclusion of the argument based on the meaning of ‘I’ is still missing a comprehensive justification in the current literature. And I guess you will just have to trust me with this due to the general complications of this type of task.

Double Reflexivity

When we think about us using the first-person concept, we are aware that we do so and thus that we ourselves are the objects of these thought tokens. This form of self-thinking can exist in stark contrast to cases in which we use individual concepts that we usually express through the use of proper names or definite descriptions that sometimes merely happen to refer to ourselves without us knowing that they do (Castañeda 1966; Lewis 1979; Perry 1977, 1979). The difference between these types of self-thinking becomes particularly clear when considering scenarios such as the well-known messy shopper case introduced in one of John Perry’s seminal articles on indexical thought. Perry once proposed to imagine the following situation:

I once followed a trail of sugar on a supermarket floor, pushing my cart down the aisle on one side of a tall counter and back the aisle on the other, seeking the shopper with the torn sack to tell him he was making a mess. With each trip around the counter, the trail became thicker. But I seemed unable to catch up. Finally it dawned on me. I was the shopper I was trying to catch … I believed at the outset that the shopper with a torn sack was making a mess. And I was right. But I didn’t believe that I was making a mess. That seems to be something I came to believe. And when I came to believe that, I stopped following the trail around the counter, and rearranged the torn sack in my cart. My change in beliefs seems to explain my change in behavior. (Perry 1979: 3, my emphasis)

While the messy shopper was in a certain sense already aware of himself before he had a relevant first-person thought, he did not realize that his thoughts were about himself all along. Before the epiphany he was, indeed, thinking about himself in virtue of thinking about the messy shopper, but in doing so he still failed to recognize that it is he himself, as Castañeda (1966) once put it, who is that messy shopper. It was only when he came to believe something which would be adequately expressed by “I am making a mess” that this fact was also somehow part of what the shopper was aware of and which, then, led to a change in how he handled the situation. In this regard, first-person thoughts exhibit not just a simple reflexivity, but necessarily some kind of double reflexivity or a super reflexivity as Guillot (2016) has recently called this peculiarity. In those cases, it is not just that the thinking subject happens to be identical with the object of his or her thought. That very fact is also somehow part of what the subject is aware of in having the thought. The thought is thus reflexive not just in being a representation of its thinker, but insofar as the reflexive aspect itself is also reflected by the representation. In thinking first-personally we do not simply refer to ourselves, but additionally and more importantly we have an awarenessFootnote 2 that we do so.

It can appear quite complicated to give a non-circular characterization of this additional reflexive element of first-person thought, i.e. a description without reference to further first-person representations. For example: (1) It is often (perhaps most often) said that in virtue of gras** a first-person thought, one is not just thinking about oneself (reflexivity) but thinking about oneself as oneself (double reflexivity).Footnote 3 (2) Another common way to characterize this phenomenon maintains that first-person thought refers to oneself (reflexivity) but additionally includes an awareness that oneself is the referent of these thoughts or an awareness that these thoughts are about oneself (double reflexivity). Both of these descriptions make use of the reflexive form of the pronoun “one”, which is typically used to ascribe to a subject states that the subject itself would express using first-person sentences. Castañeda (1966) has called such terms as “oneself”, “himself” or “herself” quasi-indicators since they are used to depict first-person mental states from a third-person perspective, which is to say that people use these words to represent other persons’ first-personal states in discourse, as they cannot use “I” for that purpose (at least in many sentence constructions of indirect speech).Footnote 4 However, one key observation Castañeda wanted to highlight regarding these quasi-indicators was that we are only able to fully understand them if we have some idea of the linguistic and cognitive mechanisms underlying the use of genuine indicators (hence his use of “quasi”), i.e. indexicals such as “I”, “now” or “here” persons usually use to speak about themselves and their own situation. Thus, even when the characterizations of double reflexivity given above do not contain the word “I”, they still make use of those quasi-indicators which apparently can only be (fully) understood by an appeal to the first-person pronoun (or one of its cognates such as “myself” or “me”) and other genuine indicators. Characterizations like (1) and (2), therefore, seem to give circular descriptions of the distinctive awareness involved in first-person thinking as they involve further indirect (i.e. from a third-person point of view) reference to the first-person concept (as in “awareness that the thought is about oneself” or “awareness of oneself as oneself”).

Despite these issues of the characterizations discussed above, there is another promising option available to describe the distinctive awareness accompanying first-person thought. (3) One could also say – and this is probably the way I would prefer to put it – that in virtue of having a first-person thought, one is also aware of being the object of thought. This last characterization – which may be taken to involve a sentence structure with the so-called PRO,Footnote 5 i.e. “the null pronominal element that […] acts as the syntactic subject of infinitives and gerunds” (Chierchia 1989: 15) – avoids an appeal to additional first-person representations, but nevertheless seems to be able to account for this additional awareness involved in genuine first-person thinking. This way of putting things allows that we can give an adequate description of the cognitive change occurring, for instance, in the messy shopper situation presented in Perry (1979) but without using any further first-person pronouns or quasi-indicators. Before the epiphany, the messy shopper was already thinking about himself but without the awareness of being the object of these thoughts, something he only gained later when gras** appropriate first-person thoughts. Hence, as well as being the referents of their thoughts, first-person thinkers are also aware of being the objects of their thoughts. After all, we seem to be able to provide a description of this double reflexivity involved in first-person thinking without ending up presupposing further first-person representations, i.e. representations which contain the first-person concept. Characterizing this phenomenon does not appear to be the major problem then.Footnote 6

The Meaning of the First-Person Concept

Yet, the more crucial question that arises from these observations is: How are we able to perform this double or super reflexive self-reference? The ability to grasp first-person thoughts appears to demand more than a mere understanding of the linguistic meaning of first-person pronouns, such as “I” in English, since it only specifies that these pronouns refer to the user of the pronoun, whoever that might be in a specific situation. The linguistic meaning of indexical terms in general is most commonly understood as something along the lines of David Kaplan’s concept of character. The character of an indexical expression type is most often defined as a function from specific parameters of the context of an occurrence of that expression to the propositional content expressed using it in that context. If we want to give a token-reflexive description of the character of “I”, it would be something like: “I” rigidly refers to the user or producer of a specific token of it (Kaplan 1989: 505). Here is a passage from Barwise and Perry who also take such a token-reflexive rule as the natural meaning of “I”:

Let us begin with the word “I”. A reasonable thing to say about this expression is that, whenever it is used by a speaker of English, it stands for, or designates that person. We think that this is all there is to know about the meaning of “I” in English and that it serves as a paradigm rule for meaning. (Barwise and Perry 1981: 670, my emphasis)

Hence, in order for someone to recognize himself as the referent of an occurrence of “I” in linguistic communication, he needs to have the contextual information that he is the user of the occurrence, which obviously is not provided by the linguistic meaning of the word itself.

It is not quite clear how the first-person concept is connected to the linguistic meaning of the word “I”. But most philosophers assume that the mental correlate of the word “I” shares many features, such as its meaning, with its linguistic representation. After all, this is why we call it the “first-person concept”. Bermúdez, for instance, proposes to understand the relationship between concepts and their linguistic representations in terms of his Expressibility Principle: “Any thinkable thought can in principle be linguistically expressed without residue or remainder” (Bermúdez 2017: 2). If this is true, then the study of the meaning of a linguistic representation should give conclusive hints about the meaning of the concepts they are used to express. And conversely, linguistic representations can (at least partly) be understood by investigating the actions which usually accompany them since those involve their conceptual counterparts at the level of thought. Lucy O’Brien also suggests taking the token-reflexive rule governing the usage of the first-person pronoun to be essential to the concept expressed through it: “The self-reference rule (SRR) states that ‘I’ refers to the subject who produced it. It is very plausible to think that this rule gives us the meaning of the first-person pronoun and governs correct usage of the first-person concept” (O’Brien 2007: 49). The latest review article on first-person thought proposes a very similar view on this question: “The most popular contemporary view is that de se thoughts are governed by a rule that they refer to whoever is thinking them” (Morgan and Salje 2020: 148).

Non-Conceptual Self-Consciousness

Many adherents of the claim that there is non-conceptual self-consciousness declare that only non-linguistic information given in experience can fill the gap and deliver us with the information that we are currently the thinker of specific I-thoughts. Thus, non-conceptual self-consciousness is considered to be a condition for double reflexive mentalese or linguistic self-reference. Here are just three quotations from philosophers who draw this conclusion from the outlined claimsFootnote 7:

[T]he first-person pronoun requires that the person referring to himself knows that he is the producer of the token in question. […] The point here is that the capacity for reflexive self-reference by means of the first-person pronoun presupposes the capacity to think thoughts with first-person contents, and hence cannot be deployed to explain that capacity. In other words, a degree of self-consciousness is required to master the use of the first-person pronoun. (Bermúdez 1998: 17–18, my emphasis)

To think of oneself as oneself, remember, is to think of oneself as the thinker of this very thought. An awareness of our present mental activity is thus key to the required form of super-reflexivity. There is no better way to achieve this than by exploiting the way that our present thinking is manifested to us as it unfolds: namely, through the phenomenology of intellection. (Guillot 2016: 146, my emphasis)

If I always knew something of myself via a term or referring token, there would be needed the additional (unexplained) fact of my knowing I was its producer. Therefore, it seems we each must have a kind of access to ourselves which is not via a term or referring expression […] What is the nature of this access not mediated by a reflexive term? […] Since knowledge of something as the referent of referring expressions is a kind of knowledge by description, we might think reflexive self-knowledge must be knowledge by acquaintance. (Nozick 1981: 81, my emphasis)

Hence, the central points of departure of this argument are the claims that (1) the meaning of the first-person concept always picks out the user of the concept and thus the owner of a thought containing it, and (2) for a thinker to recognize himself as the referent of the first-person concept, he has to be aware that he himself is the subject or owner of the thought containing the first-person concept. If a thinker of a first-person thought did not have the contextual information that he is the owner of the thought, he would not be able to recognize himself as the object of this kind of thought.Footnote 8 But how can a first-person thinker become aware that he is the owner of a first-person thought?

Obviously not by merely having additional thoughts about his first-person thought such as I am the owner of this thought (call this thought “t2”) since this would require him to be aware of the fact that he is the owner of this additional t2 as well. If this awareness of being the owner of t2 were another thought (call it “t3”) we would have to presuppose another ownership-thought (call it “t4”) which enables the thinker to recognize himself as the referent of t3, and so forth. So, claiming that having a first-person thought generally requires another first-person thought in which the thinker ascribes this thought to himself leads into an infinite regress of first-person thoughts which is a bad outcome, as a human person (a limited being) can possess only a finite number of thoughts.Footnote 9

That is why some philosophers assume that first-person thoughts come with an inbuilt non-conceptual self-consciousness of oneself being the owner of these states. This component of experience is also often called “me-ishness”, “for-me-ness”, “sense of ownership” or “subjective character”. Here is how Ned Block described this phenomenon back in the 1990s: “P-conscious states often seem to have a ‘me-ishness’ about them; the phenomenal content often represents the state as a ‘state of me’” (Block 1995: 235). Despite the prima facie appeal of this argument, I have some doubts about it. Here are three reasons why I believe that the conclusion some researchers draw is not fully justified by their given argumentation. After having presented these worries, I will present reasons that may help to put them aside.

3 Possible Objections to the Argument Based on the Meaning of ‘I’

  1. 1)

    Unconscious First-Person Thought and Double Reflexivity: The first issue is that first-person thought itself is quite often unconscious. And if double reflexivity is a necessary feature of first-person thought, then it presumably is in those cases unconscious as well. But if this is the case, then it seems that the awareness of being the owner of a first-person thought does not have to be conscious in order to recognize oneself as the object of those thoughts. Perhaps, this is even impossible since I cannot consciously recognize myself as the object of a thought or as the owner of a thought if I am not conscious of this thought itself. If I have no consciousness whatsoever that a specific thought exists, why should I be conscious of the fact that a specific subject is the owner or object of this thought?

    Another related worry is that it is not at all clear whether this double-reflexive awareness involved when using the first-person concept is a conscious phenomenon even if the first-person thought it accompanies itself is conscious. When someone is using the first-person concept, he may not be conscious of being object of this thought. In other words, first-person thoughts seem not necessarily (especially not for non-philosophers) to involve a conscious meta-representation depicting oneself as the object of the lower-order first-person state.Footnote 10 Therefore, the distinctive awareness that separates first-personal self-reference from third-personal self-reference very often is not something that is part of our phenomenology. But, if our awareness that we are thinking about us itself is unconscious, then one may be tempted to conclude that there is no reason to assume that an explanation of this kind of awareness needs to posit anything extravagant in our phenomenology, such as a special phenomenology of cognition (Guillot 2016) or peculiar forms of self-acquaintance (Nozick 1981).Footnote 11 Hence, the argument based on the meaning of ‘I’ may provide justification for the claim that first-person thoughts require contextual information that oneself is the thinker in order to involve double reflexivity, but since double reflexivity is itself a phenomenon often occurring at the unconscious or subpersonal level, the argument alone does not justify the claim that this contextual information has to be given consciously.

  2. 2)

    Subpersonal Language Processing: Moreover, it is clear that a lot of language processing is going on unconsciously and often we have to put so much effort into investigating the linguistic meaning of particular types of expressions that it seems highly unlikely that an understanding of it in everyday life could ever arise at the level of consciousness. Experts still disagree about how even the apparently most basic building blocks of natural languages – such as demonstratives or proper names – work. So, if this is true, then it seems very natural to be skeptical about the idea that the competent use of the first-person pronoun or first-person concept requires some conscious comprehension of how contextual information is involved. Even if we rightly believe that contextual information is needed to have first-person awareness, it is not clear why we should suppose that this information is given or processed consciously. Since the use of the first-person pronoun and concept apparently are not necessitated by a conscious understanding of the meaning and rules governing them, it may be concluded that the contextual information, which is additionally required to fix their reference, also need not be conscious.Footnote 12

  3. 3)

    Theoretical Implications: Finally, if the argument based on the meaning of ‘I’ were sound, we probably would also need to accept that other indexical or demonstrative thoughts are in need of an experiential foundation. In this case, then, we need to accept not just that there is a phenomenal me-ishness or a sense of ownership as some people have claimed, but also that there is some sort of here-ness, now-ness or that-ness given in experience. These implied assumptions might not be outright counterevidence against the assumption of non-conceptual self-consciousness, but they at least show that the argument based on the meaning of ‘I’ is not as theoretically innocent as its adherents may proclaim.

4 Reply to the Second and Third Objection

Let me first say something about the last problem which I take to be the least worrying, since there might be similar phenomena, as the mentioned me-ishness, associated with the use of other indexical concepts such as now or her. Think, for instance, about the famous specious present (James 1890) to which we would arguably refer by using “now”. Many philosophers have held the view that the present comes with a peculiar phenomenology that is very distinct compared to the experience of events which are given as being far in the past or future (cf. Dainton 2018: Section 2.6; Le Poidevin 2019: Section 4). Perceiving something as present evokes the feeling of being simultaneous with the things or events experienced in this temporal mode. This feeling gets lost when something is experienced as past. Hence, the idea of a now-ness given in experience is not alien to philosophy. And perhaps we can also find a here-ness of experience. For instance, the objects of perception are usually experienced as being here, i.e. co-located with the perception itself. When someone sees his wife coming, she is given as being somehow close to this perception of her. The objects of memories in contrast are not necessarily experienced as being here, i.e. as being co-located with the memory. Hence, remembering someone coming might not include this distinctive feel. So, there also seem to be differences in the phenomenal character between these types of mental states which are possibly best characterized by an appeal to an experienced co-localization or simultaneity. Therefore, even if the argument based on the meaning of ‘I’ has these theoretical implications, this is not reason enough to reject it.

With regard to the second problem, it cannot be denied that the complete linguistic meanings of words and rules which govern the use of concepts almost never cross the minds of their ordinary users. Perhaps this is even impossible because they might be too complex and would take up too much cognitive space in everyday situations. Yet, this is no reason to conclude that additional information which is needed to fix their reference does so too. Although conscious first-person thinking may not involve a conscious processing of the reference rule or character of “I”, it could still be the case that I have to be conscious of certain contextual parameters to become aware of the things they denote. This seems to be at least not as demanding as a conscious comprehending of linguistic rules or conceptual meanings. For example, usually I have to see the object to which a demonstration accompanying an occurrence of “this” points to, to know what this occurrence of “this” refers to. This is a fairly easy task compared to understanding how demonstratives work. And in the case of demonstratives, one apparently even needs to be conscious of the demonstratum to understand what the demonstrative is all about (while one probably has no idea of the complex linguistic and psychological mechanisms underlying the use of demonstratives). How could I otherwise become aware of the referent of a demonstrative? Hence, even if the processing of the meanings of linguistic devices or concepts usually occurs unconsciously, other information which is required to fully capture their reference could be or even has to be conscious. Thus, why not accept that conscious first-person thoughts involve a consciousness of their thinker or a mental-ownership-relation in order for them to include a full grasp of their reference? Unconscious meaning processing is at least no good reason to deny that there is such a consciousness of the thinker involved during first-person thinking.

5 From the Functional Correlates of Consciousness to a Reply to the First Objection

5.1 Preliminaries

However, in order to fully justify the conclusion that conscious (contextual) information is needed for double reflexive awareness during first-person thinking, there must be more justification. The reply to the second objection only shows that the complex linguistic and psychological mechanisms underlying indexical thought provide no reason to assume that contextual information required for these kinds of thoughts cannot be (or usually is not) conscious, but it does not show that this contextual information needs to be conscious in the case of first-person thought. An adequate answer to the first objection is therefore still lacking. One way to provide this is to appeal to the general function of consciousness. If we knew what consciousness is good for, we might also be able to say more about the function of particular parts of our phenomenology, such as the experience of being the owner of one’s thoughts. Put differently, an answer to the question of what the property of being conscious functionally adds to mental states – and especially those which encode contextual information needed for indexical thinking – besides the feeling of what it is like to be in them (Nagel 1974), should put us in a much better position to decide if conscious contextual information is actually required for genuine first-person thought or if unconscious information is already sufficient.Footnote 13

Of course, this is still a controversial and highly debated question. Arguably, it is even more complicated to solve than the question of whether or not there is non-conceptual self-consciousness. And my intention for the remainder of this section is not to put forward some functionalist identity theory of the mental. However, I believe that the presence of consciousness empirically correlates with a particular set of functions for actual (but perhaps not for all possible or conceivable forms of) human behavior.Footnote 14 Accordingly, although I admit that there might be conceivable differences between what philosophers call Access and Phenomenal Consciousness, I still believe that there actually is a strong correlation between consciousness and particular things we can do. In this respect, I will follow Ned Block’s proposal from 1995:

[C]ases of A-consciousness without P-consciousness, such as my “superblindsight patient”, do not appear to exist. […] This suggests an intimate relation between A-consciousness and P-consciousness […] [P]erhaps P-consciousness and A-consciousness amount to much the same thing empirically even though they differ conceptually, in which case P-consciousness would also have the aforementioned function. Perhaps the two are so intertwined that there is no empirical sense to the idea of one without the other. (Block 1995: 242, my emphasis)

After a brief (and tremendously unsatisfying) discussion of the question What is consciousness good for? and some recent attempts to answer it, I will return to the first of the noted objections to the argument based on the meaning of ‘I’.

5.2 Some Clues from the Debate about the Functional Correlates of Consciousness

It seems very natural to suppose that consciousness is closely related to human action and decision making. If someone’s consciousness is absent, it is more likely that they will not be able to perform specific actions (or at least we do not take these persons to be responsible for their actions). However, many cognitive and behavioral tasks that people thought to be only possible when consciousness is present have turned out not to depend so much on conscious experience after all. Psychopathologies in particular, such as blindsight (Weiskrantz 1986), have played a crucial role in discovering this fact. Moreover, ever since the Libet experiment (Libet et al. 1983; Libet 1985) and more recently the discovery of choice blindness (Johansson et al. 2005, 2008), doubt has been cast on the close connection between decision making and consciousness. The results from these studies suggest that the functional link between consciousness and the performance of specific actions is not as strong as we have long thought, since many (if not most) actions seem to be determined at the unconscious level.

Other cognitive functions that have been proposed as candidates for the specific role of consciousness are the integration of informationFootnote 15 (Tononi 2008) or being the property which enables the transfer of information between a wide range of different types of cognitive processes (such as perceiving, inferring, evaluating or the planning of actions) (Baars 1988). However, it also seems that information integration and the global broadcasting of information are not dependent on the presence of consciousness, since many unconscious tasks involve the integration of information and the global availability of specific types of information. That is why consciousness possibly does not play a necessary role in those kinds of activities.Footnote 16 But if this is not what consciousness is needed for, what is it then?

Given these negative results, Chris Frith and Thomas Metzinger, by drawing an inference to the best explanation, have recently suggested that consciousness might not have been primarily “invented” by evolution as something that merely enhances the success of one particular individual, but instead was brought into existence as a social tool that enhances the evolutionary success of a group of human beings. Frith and Metzinger link the role of consciousness to central aspects of human interaction – such as responsibility, empathy, modal competence or regret – and claim that “conscious content may have played a decisive role in the emergence and stabilization of complex societies. This is one prime example for a function of consciousness” (Frith and Metzinger 2016: 200). They assume that at the heart of consciousness lies what they call a “sense of agency”.Footnote 17 This concept denotes the experience of being in control of one’s own (mental) activities and is supposed to be an essential prerequisite for complex cultural developments:

Whether or not the conscious experience of agency (being in control) has any well-circumscribed causal role in the action currently being performed, the experience has an important role in culture. For example, verbal reports about this specific type of phenomenal experience can now become “theory-contaminated” and begin to drive cultural evolution. (Frith and Metzinger 2016: 204, my emphasis)

But for consciousness and its sense of agency to drive cultural evolution, as Frith and Metzinger claim, they have to play a central role within verbal and non-verbal human communication. And, as Frith and Metzinger propose, this role may be to make specific kinds of information available for communication. In this regard, consciousness seems to provide a pre-theoretical understanding of such mental entities as pains, perceptions or beliefs and the agents or owners to whom they belong, which only then can take part in our communicative endeavors. In other words, consciousness gives us cognitive access to an ontological realm (i.e. the mental) that we would otherwise not be able to communicate about. This seems like a reasonable claim since, for instance, arguably we would not have our common notion of a pain, had no one ever experienced this distinctive feeling associated with it. In fact, it seems even difficult to imagine how someone could grasp genuine thoughts about pains if they had never been able to stand in some experiential acquaintance with these kinds of things.

And if we assume that consciousness has the (primary) function of enabling information about the mental for interaction (amongst others verbal interaction), then this might speak in favor of the assumption that the conscious experience of being the thinker of one’s thoughts also plays an essential role for first-person thought and the subsequent communication of it. That is to say, we might only be able to entertain first-person thoughts, which in virtue of their meaning refer to their thinker or owner, if we can have the experience that they are being owned or thought by a subject or a thinker. Just as we might only be able to grasp the concept of pain if we can have experiences of pains, so we might only be able to grasp the concept underlying the use of “I” if we have some prior experiential pre-theoretical grasp of what a thinker or a subject of experience is. Of course, this is more of a suggestion rather than a conclusive argument. Yet, Frith and Metzinger’s theory might give proponents of the argument based on the meaning of ‘I’ at least some abductively justified hope.

Let me briefly present another recent social neuroscientific theory of consciousness, which suggests that consciousness is primarily a tool for communication. Graziano (2013, 2017) and Graziano and Kastner (2011) claim that the same cognitive abilities and neural mechanisms forming the base of social cognition are at play when someone gains consciousness. They assume that consciousness is a by-product of higher-order perception of one’s own awareness which works similarly to the perception of other people’s awareness.Footnote 18 The difference, however, is that the awareness in both kinds of cases is experienced as coming from different locations (ibid.: 101). Even though I do not agree with this higher-order approach to consciousness, they also propose something that I think is quite right, namely that “much more can be learned about consciousness by […] asking what, specifically, consciousness can do in the world, what it can affect” (ibid.: 112, emphasis in the original).Footnote 19 According to them, consciousness does not only emerge because information is, in the right sort of ways, processed, it certainly has an impact on information processing as well. Graziano claims, like Frith and Metzinger, that one of the central functions of consciousness is to provide information about the mental for communication. He writes for instance:

Try explaining colors to a congenitally blind person. (I actually tried this when I was about fourteen and lacked social tact. The conversation went in circles until I realized he did not have the concepts even to engage in the conversation, and I gave up.) However, as limited as human language is at information transfer, and as indescribable as some conscious experiences seem to be, we can nonetheless report that we have them. Consciousness can affect speech. It is tautologically true that all aspects of consciousness that we admit to having, that we report having, that we tell ourselves that we have, that philosophers and scientists write about having, that poets wax poetic about, can have some impact on language. (Graziano 2013: 41)Footnote 20

Again, if it is true that consciousness makes information about mental entities available for communication and (as a prerequisite for this) for thinking, then it seems that in order to be aware that oneself is the thinker of a first-person thought, one has to have some prior understanding of what thinkers, thoughts and their relations are, concepts which might only be gained by being conscious of such things these concepts are about. Therefore, at least at some point we may need to have had conscious thoughts to gain these concepts so that they can as such play a role in further thinking and communication.

5.3 Reply to the First Objection

After having explored these ideas about the function of consciousness, let’s finally address more directly the first and presumably most serious worry presented in Section 3. It is probably true that there are unconscious first-person thoughts and that in those cases we actually have no consciousness of being the object of these thoughts. However, since they are first-person thoughts, this double reflexivity discussed in Section 2 has to be part of the representational structure of those mental states as well. So, it may not be the case that one is always conscious of being the object of one’s own first-person thoughts, but in virtue of entertaining this kind of thought, one is nonetheless (unconsciously) presented as such. Similarly, mastery and an adequate understanding of the token-reflexive rule governing the first-person concept also requires mastery of such concepts as thought and its thinker. And I agree that we probably only need to have states which unconsciously present oneself as the thinker of these states to be able to entertain unconscious first-person thought. However, presenting oneself as the thinker of thoughts or understanding the meaning of the first-person concept involves an understanding of such concepts as thought and thinker and of the concepts of the relations holding between both. These are mentalese concepts which we are only able to grasp if we have some prior experiential access to the entities they denote (if it is true for what I have argued before in Section 5.2, namely that only consciousness enables one to think about one’s mental states).Footnote 21 And because even unconscious first-person thought seems to depend on these mentalese concepts, one has to have some prior experiential access to one’s own thoughts and oneself as their thinker. That is why even unconscious first-person thinking seems to require something like a subjective character given in experience at some point prior to its development.

6 Conclusion

I hope I have been able to show that many philosophers who argue for non-conceptual self-consciousness on the grounds of the outlined argument owe us more justification than they usually offer, but that there are nonetheless attractive theoretical possibilities to get their argument to work. I agree that an understanding of the meaning of the first-person concept alone will not suffice to recognize oneself as the referent of an occurrence of the concept. One also has to be presented as the thinker of a first-person thought to become aware that those thoughts are about oneself. And I also agree that this thinker-presentation, on the other hand, cannot be in each case a thought containing the first-person concept itself since this would lead into a regress of mental states. However, the claim that this non-conceptual thinker-presentation needs to be conscious is much more controversial in my opinion and needs additional justification. To that end, I gave some argumentation as to why this might be the case. Representing oneself as the thinker of a thought or understanding the token-reflexive meaning of the first-person concept seem to entail the mastery of such concepts as thinker or owner of a thought. Yet, these mentalese concepts might only be gained by having some prior conscious access to the entities, i.e. thinkers, thoughts and their relations, they denote. Just as a person who is congenitally insensitive to pain might not be able to grasp our common concept of pain, so a person, who has never consciously experienced thinking, is presumably incapable of gras** the concepts which adequately represent the entities involved in thinking and thus the concept necessary to fully grasp the reference of the first-person concept. Hence, there might be very good reasons to assume, as many (and perhaps even the majority of) philosophers working on these issues do, that there are non-conceptual or non-linguistic forms of self-consciousness. The metaphysics of these forms of self-consciousness is nowadays discussed using such terms as “subjective character”, “pre-reflective self-consciousness”, “me-ishness”, “for-me-ness” or “mineness”.Footnote 22