Blog Coordinator

« The True Self and Reflective Endorsement | Main | Racial Bias and the True Self »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Josh: So let's bring the two posts thus far together. True self theorists (in the philosophical tradition) try to identify which particular subset of mental states is expressive of the agent's "true self." You object to the mental states approach in favor of a value-based approach to identifying essence. But isn't there room for a combined approach? That is to say, what if we go ecumenical in the determination of the relevant mental states that are or can be where the true self is located? They could be desires, beliefs, cares, OR judgments say. But this collection is nevertheless still a *subset* of the agent's mental states, ones that have a privileged place in the determination of the agent's true self. Not all mental states, after all, are going to be expressive of the agent's true self, e.g., visual perceptions or pains will be excluded on any account, I take it. So perhaps the lesson to be learned here is that there's reason to combine a mental states account and a value-based account? That is to say, WHICH mental states (of the privileged subset) in any given case count as expressive of the true self is going to be a function of the assessor's normative stance.

Thanks for this post (I look forward to more!). In the previous one, I asked if you thought the values involved in determining what (who?) is the true self were mainly internal, external, or a combination. This post seems to move to emphasize external considerations of value, which is natural given part of the motivation is empirical data. But I have two questions. One, could the emphasis on external evaluation stray too far toward stereotyping essences of people? Two, when external evaluation of one's essence is at odds with what an individual evaluates as one's essence, is there some way we could decide who's right?

Hi David,

This is really a great point. What I meant originally is that we won't be able to make progress on understanding the notion of a true self by drawing distinctions between different types of mental states (e.g., by distinguishing between desires, beliefs, cares, and judgments). Still, one may say that, at the very least, the only things that get to be part of the true self are the agent's mental states.

I am not sure. Suppose that a child grows up exposed only to vapid pop music. She enjoys listening to this stuff and has never considered any other option. Then, one day, she encounters a punk rock for the first time. She immediately feels that this music really speaks to her. So she stops listening to that pop and becomes a punk rock fan.

Now consider the time before she was exposed to punk rock. At that time, she did have a mental state of liking pop but did not have a mental state of liking punk. Still, I would say that listening to punk would have been a more genuine expression of the true self she had even at that time.

In other words, my intuition is that the true self she had at that time was not exhausted by the mental states she had at that time. Does that sound right to you?

Hi Alan,

Good questions. I don't know if you will find this answer helpful, but the idea is basically that we can answer questions about the true self in just the same way we answer other questions about essence.

For example, suppose that Democrats and Republicans have a disagreement about the essence of the United States. Democrats say that Donald Trump's anti-immigrant rhetoric goes against the essence of what our nation is all about; Republicans disagree. It might be hard to know precisely how to adjudicate such a disagreement, but my thought is that whatever we would do in adjudicating a disagreement like that, precisely the same approach would make sense for disagreements about the true self.

Hi Josh,

Thanks for your reply to my previous comment. I'll follow it up here, as what I now want to ask about has something to do with this second post.

In your reply to me in the other thread you considered what people might think about a wanton person. Now, I must admit that I find it very difficult to wrap my head around this example. It seems to me distinctive of persons that we can and do have reflective attitudes--we have attitudes about our attitudes. This is part of why I shied away from person talk in posing my question. It seems to me that a wanton is in a crucial respect not like a person.

In this second post, you mention a general strategy for thinking about the true self that applies to more than just persons--e.g., articles, countries. I see that this essentialism is not narrowly focused just on issues related to personhood. But I wonder if there is reason to prefer something like the ecumenical strategy that Dave suggests in his comment when it comes, specifically, to the true SELF.

My thought is this: Perhaps the essentialist thinking is correct, in general, but there is something distinctive about persons, such that the true self, in particular, is best conceived of in terms of mental states (perhaps among other things). The distinctive feature of persons is that we have reflective attitudes. One without these attitudes just wouldn't be the sort of creature with a true self to begin with. But, as Dave suggests, the normative considerations may come into play when determining which of the agent's attitudes are constitutive of the true self.

I wonder what you (and others) think.

Thanks, Josh. I don't have much to add except:

1. I agree 100% with your approach here (for whatever that is worth!).

2. I think agency scholarship can benefit from intersection with other areas of philosophy. Here, you invite consideration of work on essentialism, which is an entirely independent field that I know nothing about. Another ripe candidate is work on Rawl's Veil of Ignorance.

There's also a link between Rawl's argument and your work on essentialism here, because Rawl's asks us (arguably) to imagine ourselves as other people (in some sense). If our essential selves are small enough to be common with other, random strangers, then this experiment makes sense. If our essential selves are thicker and more brittle, then the argument makes less sense (and, one can argue, is incoherent). This is why I say that the argument from constitive luck is the strongest argument for free will antirealism, and the best counter-argument is that the concept of constitutive luck is, in some important way, incoherent.

Dennett parodies the incompatibilist argument by alleging that, according to incompatibilists, the essential self is so small that it is just a mathematical point, practically blank. I think he is exaggerating, but only slightly. At least one part of our moral psychology thinks of human beings as highly malleable, with very small essential selves, and this part supports the free will antirealist argument. But, I suspect, our natural moral psychology is probably more complicated than that, with multiple competing intuitions, and some of those leading toward thicker views of essential selves.

3. I mentioned this before, but I didn't get a clear answer, so I'll add another gentle reminder. How does this work on essentialism tie in with Neil Levy's work on constitutive luck? Or is that getting too far ahead of ourselves - first, let's see how essentialism works, and then we'll see how that impacts the free will debate?

4. It's also possible that our concept of an essential self is fundamentally incoherent. It appears, superficially, to make sense, but scrutiny of border cases shows that there is no real essential self. In other words, the essential self is a linguistic crutch that we use, and it works in 99.9% of situations, because most people don't undergo radical transformations most of the time, leading to embarrassing border cases where the term doesn't clearly apply or not apply. If the concept of an essential self turns out to be an illusion (and I'm not sure it will, but I think there are strong arguments that it will), I'm not sure what the impact would be on free will scholarship. You could argue that free will doesn't exist, in that case, but we wouldn't reach that conclusion for the normal reasons (e.g., reasons like the manipulation argument). The debunking would go even deeper than that. It would be more like Tamler Sommer's meta-skepticism about moral responsibility.

Hi Ben,

Thanks so much for your continued engagement. I'm thinking that maybe the best way to get to the bottom of this would be for me to try to state a little bit more clearly the position I was trying to develop and then for you to save a little bit about where you think it is going wrong. Okay, here goes...

First, consider a band. We can think of the band as a collection of members (a guitarist, a drummer, etc.), but we might also think that there is something like an *essence* of the band. That is, we might think that some of the band's albums are genuine expressions of what the band is really all about and that others are best understood as betraying that essence.

A key point here is that the essence of a band is not something like a privileged subset of the members of the band. Rather, when you are thinking in this essentialist way about the band, you are not just thinking of it as a collection of members. You are using a very different kind of a conceptual framework.

Now consider the way we think about the self. We can think of the self as a collection of mental states (beliefs, desires, etc.), but we can also that there is something like a *true self*. That is, we can think some of your actions our genuine expressions of who you really are and others are best understood as betraying the person you really are.

What I was trying to suggest is that these notions cannot be spelled out in terms of the framework philosophers have developed for thinking about collections of mental states. The true self is something like a privileged subset of the agent's mental states. Rather, it is best understood in a completely different way. Instead of thinking about collections of mental states and the properties they have, we might start by just trying to understand what people mean when they attribute essence to these other entities. The true self could that be understood against the backdrop of that very different sort of conceptual framework.


Just wanted to say that I appreciate these thoughts about constitutive luck and and I'm looking forward to thinking more about them. As I said earlier, it seems clear that people find the problem of free will to be confusing. Clearly, we need some way to explain that confusion, and this sounds like a very plausible candidate.

Hi Joshua

Interesting posts. On the kind of view you're heading towards, is it possible for a person's "true self" to change over time? In my view, it's quite reasonable for a person to have a particular "true self" at t1 and a different (even if only slightly different) "true self" at t2. So my love of vapid pop music might have been expressive of my true self earlier in life, but my love of punk music might be expressive of my true self later in life. My true self simply changed between t1 and t2.

So I see where you're going with this, Josh, and it's (as always) very interesting and insightful. So in response to your earlier response to me, but also to move things forward, I'm trying to think about how philosophers steeped in the "mental states" tradition are going to continue to reply to you and how you might convince them otherwise. So in your vapid pop -> punk lover case, I'm frankly not sure what to think, but if I allow that listening to punk would've been a better expression of her true self during her vapid pop days, this could easily be put in terms of a disposition, perhaps an emotional disposition to be aesthetically moved by punk rock, where this could be put in terms of non-rational cares. Then it was a care (mental state) of hers during her vapid pop years, just one about which she wasn't aware and hadn't yet been activated (and I don't think this sort of thing is at all rare -- there are plenty of events I have had very surprising emotional reactions to).

Now whether or not this is a plausible analysis of the case, how might you move someone off of it in favor of a more, I don't know, holistic view of the essential self (which is the analogue, I take it, of the band analogy you give)?

Hi Josh, great stuff as always! Unless you plan to discuss it later, can you say what you think of the sketch of a theory I've discussed with you before? Namely: People are "proper function essentialists," at least about categories of things that are taken to have a proper function, including bands, nations, scientific papers, and persons. The essence of these things is to function as they are supposed to. Hence, when we make judgments about whether they are expressing their essence (or "true self") more or less after undergoing a change, or in behaving in one way rather than another, we are deploying our (implicit) theory of what the thing's proper function is.

For instance, a person's proper function is to flourish (as a human), so judgments or behaviors that conflict with flourishing do not express one's true self--from taking heroin to suppressing one's sexuality (from a liberal's perspective) to failing to reflect on one's desires (being a wanton), and perhaps also being cowardly, getting a disease, or being insensitive to a friend. Similarly, a band's proper function is to play good music, so it is less itself when it moves away from doing that. And same with scientific papers, etc.

This view is consistent with judgments that some (non-typical) things have proper functioning that involves morally bad behaviors. So, we might see some interesting judgments about whether immoral behaviors and beliefs express the true self of Hitler or Charles Manson, about whether a band formed to create crappy pop expresses their true self when they create musically sophisticated songs, whether a now dulled machete is more or less itself when it fails to cut off an arm, etc.

My own (liberal) intuition is that the thoughtful remarks made by Trump in response to America's recent tragedies are less expressive of his true self than the insensitive invective that typically spews out of his mouth (though he's a particularly tricky case to consider when thinking about 'deep' selves!)


You said, "Quite generally, when people are thinking about the essence of an entity, they tend to regard the good aspects of that entity as more essential."

Do you say "Quite generally" here because in some cases people regard the bad aspects as essential to an entity? For instance, think of a deeply religious person's thinking about the essence of the devil, or Harry Potter's thinking about the essence of Lord Voldemort. When these people think about the essence of these entities, I would think that they regard the bad aspects of that entity as more essential.

Hi Eddy!

This is a great suggestion. Though I wouldn't want to fully commit myself to it, I definitely think that it is very likely to be correct.

Here are a few things that speak in favor of your hypothesis:

1. It is completely general, applying not only to human beings but also to non-human entities like scientific papers, bands and nations. Thus, if we go with your approach, we don't have to come up with some special account about human beings in particular. The account for human beings just falls out of this more general theory.

2. It helps to explain why we should be especially likely to see people's true selves as *morally* good rather than as good in some other way. For example, when I do things that are morally bad, you might think that my true self is calling me to behave otherwise, whereas if I do a terrible job playing basketball, you would be unlikely to think that, deep down, my true self was calling me to make different basketball decisions. This would fall out of your hypothesis very naturally, if we just assume that it is part of my proper function -- my telos -- to behave morally but not to play basketball well.

3. It helps to explain how it could sometimes be possible for an agent to have a morally bad true self. In ordinary cases where a person does something morally reprehensible, we assume that she is going against her own telos. However, it is at least possible to imagine someone (Matt gives the example of Voldemort) whose telos is actually to be evil.


Exactly! People seem to have a very strong tendency to suppose that the true self is morally good, but it also seems that this tendency is defeasible. In our experiments on this issue, we have tried to manipulate this in a number of ways. First, we tried just telling people that the agent did lots of bad things, but in that case they still assumed that the agent's true self was good. Then we tried explicitly telling people that the agent had a morally bad true self. This did have some impact, though we were never able to get people to regard that agent's true self as being nearly as bad as they would regard, say, your true self as being morally good.

(Just in case you're curious, this is all in Experiment 5 in the paper at:

Hi David,

That is a great point. Just as you say, it would also be possible to accommodate this result on the mental state theory using exactly the approach you propose.

I guess that my views on this topic are motivated in part by an attempt to make sense of the more general phenomenon. In thinking about the essence of a scientific paper, it seems that we don't want to appeal to the paper's mental states (just because it doesn't have any). Thus, in that type of case, it seems like we have to look to something more holistic. Then my thought is that if are going to be looking to something more holistic in these other cases, we should adopt the same approach when thinking about the true self.

That said, I certainly agree that this argument is far from conclusive. Perhaps the best way to resolve the matter would be not just to think at this more abstract level but instead to focus on one specific way of spelling out the more holistic approach (say, the one Eddy proposes above) and one specific way of spelling out the mental state approach (say, yours). It would then be far more possible to get into the nitty-gritty of whether each of these approaches could actually be made to work.


That's a really helpful question. In itself, I don't think my approach would be committed one way or another on the question as to whether people think the true self can change. Rather, the approach just predicts that people should think the same thing about the true self that they do about the essences of other entities. So if people think there can be changes over time in the essences of other entities (nations, bands, disciplines, etc.), the approach predicts that they should also think that there can be changes over time in the true self. Conversely, if people think that the essences of these other entities are immutable, the approach predicts that they should also think that the true self is immutable.

We have not done any work on this question ourselves, but there is a really nice paper on it by Christy, Cimpian and Schlegel. They suggest that people show a tendency to think of the essences of these other entities as immutable, and they then show that people also tend to think of the true self as immutable.

But of course, this is just a statistical tendency. It's not that no one ever thinks the true self can change in any way. It's just that people show a far greater tendency to think that your true self will not change than to think that other aspects of you will not change.

Hi Joshua,

As before, that's very interesting. Thanks for posting.
With regard to objects such as papers, or institutions, etc., it seems to me that any agent that competently uses language and can name a paper, or an institution, etc., also needs to learn some conditions under which the term is applicable. Those conditions might be fuzzy, underdetermined (i.e., determined only up to a point until further information is given), etc., but still, some conditions are needed for proficient usage of a term, in this case a proper name.

In light of that, any such agent (human or otherwise) who is asked whether after certain changes, the paper, institution, etc., is the same or was replaced by another, and is not given the option to elaborate, would (probably) properly reply that some changes would result in the replacement of the paper, institution, etc., and others would not.

However, that might not be the result of essentialist thinking with regard to those things, but of concept acquisition and mastery (unless you're using the idea of essences in a very ontologically-light way).

So, my question would be: Have you done some experiment to distinguish between those two hypothesis? (i.e., they're thinking about essences of things like papers, etc., as well, or they're just learning concepts?).

Or do you think that use of language requires essentialist thinking all around (but if so, I'd be inclined to say the requirement would be very light, from an ontological perspective).

Hi Angra,

Thanks for these further thoughts. In one of our experiments, the experimental participants themselves are actually asked a question using the word 'essence,' so we do have good evidence for the claim that people are thinking about these things as essences.

Still, a question remains about how people understand these essences from an ontological perspective. I don't know the answer to that question, but my guess is that ordinary folks just have no views at all on it. (This hypothesis draws on and approach that Eddy, writing in another context, refers to as 'theory lite.') In other words, my guess is that people have beliefs about the essences of various specific papers, nations, etc. but that people don't have any beliefs at all about the ontological status of the essences they are attributing.

Hi Josh,

Thanks for this further reply.

It actually seems to me as if the band analogy you bring up supports the claim that we should temper the conclusion you want to draw from these studies. But perhaps I'm missing something. So let me try to spell out my thinking and then get your reaction to it.

The analogy seems to go like this. The band is like the person, and the members of the band are like the mental states traditional theorists take to be relevant to our conception of the true self. Let's grant that "the essence of a band is not something like a privileged subset of the members of the band. Rather, when you are thinking in this essentialist way about the band, you are not just thinking of it as a collection of members. You are using a very different kind of a conceptual framework." What I still don't see is how this shows what you took it to show in your original post, namely, that reflective mental states don't have anything to do with our conception of the true self.

Here is why. It seems to me that if there were no members, then there would be no band. That is, the existence of the very entity to which we are ascribing an essence depends, in part, on the existence of members. (The particular individuals who are the members of the band may change to greater or lesser degrees over time, but there still need to be band members to have a band. Or so it seems to me.) By analogy, we might say that if there were no relevant mental states, then there would be no person. And so the existence of the very entity to which we are ascribing a true self depends, in part, on the existence of these mental states.

Thus, it seems to me that we are warranted in drawing a more tempered conclusion from your studies. Perhaps our conception of the true self is not wholly determined by considerations pertaining to reflective attitudes, because this conception is also dependent on normative considerations. However, it seems that our conception does depend on these attitudes in the sense that there would not even be an entity about which to conceive of it as having a true self if the attitudes were not present. (It may depend on these attitudes in further ways as well, but it seems clear to me that it depends on them at the very least in this way.)

Something like this thought was behind my earlier comment regarding "wanton persons." And it seems correct to me--both that we shouldn't abandon the relevance of reflective attitudes in thinking about the true self and that we shouldn't ignore the relevance of normative considerations (you have, I think, convinced me of this latter point!).

But I wonder what you think. Do you still see no role, however tempered, for reflective attitudes in our thinking about the true self?

Hi Ben,

This idea sounds intriguing, and I would love to hear more about it. Putting your point into the framework I have been proposing here, I wonder if we might describe it as follows:

People can attribute essences to entities of many different types (human beings, scientific papers, bands, etc.). However, not all of these entities have true selves. The essence of an entity only counts as a true self when that entity has a self. So when we are trying to determine whether something belongs to an entity's true self, we need to determine not only (a) whether that thing belongs to the entity's essence but also (b) whether the entity even has a self at all. It is with regard to this second question that second-order states play a role.

Does all of that sound right to you?

Hi Josh,

Yes, this does sound like what I was suggesting. Now, I wonder whether you think it's plausible, especially in light of your work on the topic. Is this suggestion compatible with the studies you've done?

Hi Ben,

Okay, good. I feel like I now have a much better understanding of the key idea here. This idea certainly sounds like a plausible one, but unfortunately, I'm afraid my work on this topic doesn't really shed any helpful light on it.

The claim I was trying to argue for was just that second-order states play no role in the distinction people draw between the true self and other parts of the self. However, just as you suggest, this claim is completely compatible with the view that second-order states do play an important role in the distinction between entities that have a self and entities that do not have a self at all.

Hi Josh,

This is very interesting stuff! I wonder whether you think it's plausible that the tendency you observe (i.e, to regard the good aspects of an entity as more essential) should be understood as a type of positivity bias. If so, then we might expect that, in some cases at least, overcoming this bias would deliver a more accurate view of the essence of the entity in question. I think we engage in such bias-overcoming in thinking about our own true selves. For instance, I might be filled with nationalistic pride upon learning about my country's military success. And, in spite of this, having recently been convinced of the truth of pacifism, I might judge my nation's military activity to be objectionable. (This example I borrow from Helm.) Now, suppose I display the tendency you note, and take my (good) pacifist attitudes to be expressive of the true me. 'I'm essentially a pacifist' I think approvingly. If my pattern of emotions nonetheless continues to express care for pro-military nationalism, I might come to doubt that I really am a pacifist. This might be a source of shame, and more importantly, serve as an occasion to reflect upon the deep-rootedness of my nationalistic upbringing. Although my self-evaluation thereby diminishes, I think there's a strong intuition that I acquire a better view of my true self in coming to appreciate that I have a lot of work to do before I really am a pacifist.

All of this is consistent with your claim that "Quite generally, when people are thinking about the essence of an entity, they tend to regard the good aspects of that entity as more essential." But, two things follow if the above thought is right. First, exceptions to the general tendency will not only include certain entities (e.g. Satan, Voldemort), but certain aspects of given entities. Particularly, certain 'deep-rooted & bad' aspects of myself might come to be immune to the tendency (even if I continue to display the tendency generally in thinking about my true self). Secondly, if this is right, and the tendency is a sort of bias to sometimes be overcome, then we must have recourse to other criteria for picking out essences (at least, when it comes to selves).


This is super helpful point, which brings up a really interesting alternative hypothesis. My own view is that the effect is not due to positivity bias, so maybe I'll just explain how I'm understanding it and then see whether you agree.

Suppose we say something like: 'Donald Trump's anti-immigrant rhetoric goes against the very essence of what our nation is all about.' In speaking this way about the essence of our nation, we don't seem to be making a merely statistical or descriptive claim. (We are not simply saying that this rhetoric goes against the majority of the things our nation has done, or something of that sort.) Rather, it seems that claims like this one are fundamentally value-laden. When we argue about the essence of our nation, the argument truly is one that involves questions of value.

Ok, with that in the background, consider again your case of the would-be pacifist. We can now consider two different claims he might make about himself.

First, suppose he thinks: 'The majority of my emotional reactions are pacifist ones.' This seems to be something like a straightforward scientific hypothesis he is making about himself. If it turns out that this belief is actually driven by value judgments, then I would agree that he is falling prey to a bias.

But now consider a belief like: 'Ultimately, these pacifist emotions express my true self.' My suggestion is that this belief isn't something like a scientific hypothesis. Rather, it is more like a belief about the essence of the United States. Thus, if his value judgments play a role in this belief, that role need not be a bias. The question under discussion here simply is one that involves questions of value.

Thank, Josh. That's helpful. I think I agree with at least half of what you say.

I'm on board with the idea that essence-talk is value-laden. So, in saying 'anti-immigrant rhetoric goes against the very essence of our nation', right, the claim made is not descriptive/statistical, but one that invokes something like "the ideal America".

My objection, though, is something like this: in saying that attitudes x but not attitudes y are part of my "true self", I thought I was saying that attitudes x (but not y) are attributable to me, and thus, are something for which I can be praised or blamed. I wouldn't say that statements about such attitudes are purely descriptive, but nor would I say that they are idealizations. So, I grant that I'm not making a descriptive claim when I say "pacifist attitudes are essential to me"; I'm pointing to something like an idealized self, and I think it's interesting that essence-talk is idealized in this way. But, do I deserve praise for my idealizations? I don't see how we could justify the idea that pacifist attitudes (x) are 'more attributable' to me than nationalistic attitudes (y). Suppose I think that my attitudes x are praiseworthy and attitudes y are blameworthy. This thought is value-laden in the sense that it expresses my evaluation about what's praiseworthy & blameworthy, but it refrains for giving me credit for what I am still trying to become.

I suppose, then, I'm wondering whether you're suggesting that we separate the concept "true self" either from attributability or from praise and blame. Either separation, I think, would be revisionary, and while it would perhaps be question-begging of me to say that the true self *just is* something to which (praiseworthy/blameworthy/neutral) attitudes are attributable, I wonder whether it wouldn't be more theoretically useful to posit something like an "ideal self" which essence-talk tracks in the case of persons. One advantage to doing so is that we commonly identify (and try to minimize) the gaps between our actual and ideal selves (and between ideal and actual instances of various types of entities).

The comments to this entry are closed.