top of page

"The Concept of Concept": A Conversation with Peter Wolfendale (Keywords: Truth; Reality; Cognition)

White house on hill

From The Philosopher, vol. 109, no. 3 ("Concept and Reality").

If you enjoy reading this, please consider becoming a patron or making a small donation.

We are unfunded and your support is greatly appreciated.

Anthony Morgan (AM): The relationship between concept and reality is an age-old concern of philosophers, and there are numerous questions to be asked when thinking about it. What do you think are some of the most interesting and/or pressing of these?

Peter Wolfendale (PW): The obvious philosophical response is of course to answer the question with a question! So, what do we mean by concept and what do we mean by reality? And this is where all of the controversies really begin. The concept of “concept” is one of the most fraught in the whole philosophical tradition, not just between analytic and continental schools but internally to each and even in relation to other disciplines such as psychology. It’s a serious source of people talking past one another. And this is because what we ordinarily use the language of concepts to do isn’t necessarily what philosophers are paying attention to when they’re trying to generate a theory of concepts. Mark Wilson would say that if you want to get a grip on the concept of concept, you’ve got to start with the ways in which we evaluate conceptual competence. So we might say, “Timmy has grasped the concepts of calculus”, and this means that he has progressed through various different mathematics courses to the point at which he has a decent understanding of what differentiation and integration are, and can perform the relevant tasks that are associated with them. But Timmy’s grasp still isn’t as good as, say, Samantha who is doing a university level course – or someone else who is doing a PhD in mathematical analysis. Wilson thinks that these evaluations are much more context-sensitive than we tend to think, shifting not just with educational level but with disciplinary boundaries and their practical concerns. What counts as mastery of the concept “toughness” varies depending on whether one is a working baker, an industrial engineer, or a material scientist. A single concept can bundle together a surprising variety of concerns, and we must be careful not to collapse them for the sake of convenience.

What we ordinarily use the language of concepts to do isn’t necessarily what philosophers are paying attention to when they’re trying to generate a theory of concepts.

AM: It’s one thing to be competent at something like calculus, but how do you become competent at some of the big nebulous philosophical concepts? How do you become “competent” in reality or in truth or something like that?

PW: It may seem like a bit of a detour, but how about we go back to Kant? There’s a familiar story that says that Kant unifies two different strands within epistemology coming out of rationalism and empiricism. And at this point the actual vocabulary that’s being used is still fairly loose and up for grabs, so people are sometimes talking about ideas, and sometimes talking about concepts – both almost interchangeably. But the key question for them is the relationship between thought and sensation. The rough idea is that the rationalists try and explain thought in terms of sensation – they see sensation as a kind of weak form of thought. The empiricists do the opposite – they see abstract cognition as some derivative of sensory impressions. Kant comes along and says, “No, you need both of these distinct things”. And indeed the question is: how do these two things connect? How is it that we can go from sensory experiences of things in the world to judgments about those things that enable us to draw inferences, make arguments, and develop elaborate theoretical explanations of the things we sense?

Conceptual understanding is something more than just sensory representation and discrimination.

Partly what this does is to centre what you might call “recognition” as the key topic for understanding conception and how it works. This is the ability to classify singular things in experience as falling under general concepts. This remains the core case for most psychological theories of concepts. But at the same time this distinguishes what counts as understanding a concept from mere classificatory ability. We can train computer systems to accurately label and even generate pictures of tigers without them thereby understanding anything about tigers, and we can teach blind biologists what tigers are even if they will never be able to visually classify or imagine them. Conceptual understanding is something more than just sensory representation and discrimination. It concerns the implications of classification: knowing that the recognised tiger is therefore a mammal, a carnivore, a potential threat, and so on. This is more obvious when we consider concepts that are further removed from immediate experience, such as “electron”, “lymphocyte” or “tectonic drift”. When we get to concepts such as “reality”, “truth” or “justice”, this connection with sensory experience has been more or less completely severed, and so competence consists purely in how we reason with them. Kant borrows the term “Idea” from Plato to distinguish these abstract concepts. Whether the right pattern of reasoning can be adequately captured by a definition and what it means for there to be a right one is a problem the philosophical tradition has been since grappling with since Socrates.

AM: So, what are the main divergences in the ways philosophers talk about concepts?

PW: The big distinction is between treating them as ideals governing cognition or as functional components of it. When I say ideals, these can be either metaphysical or social. For Plato, any talk of concepts is really just talk about universals that exist independently of human mentation (Frege holds something similar). But there are also those who see concepts as social ideals – norms or rules that have been instituted in a way that directs thought and talk. A good example of this is Hegel or someone like Robert Brandom who has well worked-out theory of concepts as norms governing the use of words.

Viewed as cognitive components, concepts are the things that we’re actually using in thinking rather than the standards against which this thinking is to be judged. And again, this can be understood either psychologically or sociologically. So, psychologists and certain philosophers will want to talk about the concept as something that is “in your head”. The idea is that you have a concept, which means you have some kind of representational structure that enables you to think about a particular type of thing in the world. This might or might not be linguistic, and thus amenable to definition, depending on who you talk to. But equally, there are others who see concepts as a sort of extended social infrastructure enabling cognition.

Take the concept of “black hole”: you and I have some kind of appreciation for what a black hole is. We can have a discussion about it within certain contexts, we might be able to talk about our favourite science fiction and how black holes are being represented in it. But fundamentally our understanding is limited and conditional upon the way in which it is embedded within the actual social institutions of physics. This is what you might call “semantic deference” – we have some grasp of the meaning of “black hole”, but there will always be points in our discussion where we’ve got to defer to those with more expertise than us. Similarly, does our use of an empirical concept like “conductivity” make any sense without reference to equipment capable of measuring it, and the wider systems through which it’s calibrated? So, even when we’re engaging in seemingly individual thought about certain topics, that cognition is actually predicated upon a network of expertise and technical apparatuses.

Are concepts doing the work of cognition, or are they the standards by which that work is to be judged?

The interesting question here is: are concepts doing the work of cognition, or are they the standards by which that work is to be judged? Here you can see a conflict in the guiding metaphors underlying concept talk: we often frame evaluations of conceptual competence in terms of possession, for example when we say that Timmy has acquired the concept of differentiation. That metaphor suggests the psychological view where we see the concept as something that is in the head. But we might equally say that Timmy has a grip on the concept, suggesting that it is something outside him, be it social or metaphysical, which can be grasped in different ways and to different degrees. There is something to each way of talking, but what we really want is to explain the relations between these different dimensions: psychological, social, and maybe even metaphysical.

AM: Could you say something more about the social nature of a concept?

PW: One way of thinking about this is to think of a concept as a practice that we’re engaged in, like a game that has got certain rules. Brandom, following Wilfrid Sellars, likes to call this “the game of giving and asking for reasons”. It’s a game in which we use words to make and trade claims about the world, making observations, drawing conclusions, and acting upon them in appropriate ways. Understanding concepts means understanding the roles that the relevant words play in this game, especially their contribution to inference, i.e., the implications I mentioned above. My individual behaviour, like the way in which I am using a word – like “tiger”, “black hole” or “justice” – is in some sense beholden to the standards set by the linguistic community. I might draw the wrong inferences, according to them. What we saw in the “black hole” case is that this authority can be explicitly codified in the form of particular institutions and experts, so that the practice incorporates deference as well as inference. But there’s a second kind of assessment that we might make, which is whether or not that practice – the ways in which things are being done standardly by the community – is correct or not. So, there’s the way in which the concept as this socially extended thing actually is functioning versus the way in which you might want to say it should function.

AM: In relation to a concept like “black hole”, how would we look at the question of how it should be functioning because obviously that’s not something that most people are going to have a say over?

PW: This is exactly where the question of the relationship between concept and reality comes in! There might be various ways in which a concept qua social practice isn’t doing its job, but the most obvious is whether or not it’s accurately representing the world: what if physicists have us all reasoning in the wrong way about black holes? We can approach this question from the other direction: why can’t they simply stipulate that they’re right? The short answer is that there is no truth without the possibility of error. The long answer requires a detour through Hegel.

In the introduction to the Phenomenology of Spirit, Hegel develops an account of what he calls “natural consciousness”, which is basically just an account of belief and/or assertion. I think it’s genuinely one of the most brilliant bits of philosophy I’ve ever read. His initial definition is very simple:

1. Consciousness relates itself to its object. For example, I may believe something about an object.

2. Consciousness distinguishes between its relating and its object. So, I recognize that the way the object is in itself could be different from the way I think of it. I acknowledge the possibility of error.

But he draws two more subtle implications from this:

3. Because consciousness itself makes this distinction, the object cannot be truly in itself, but must be for consciousness. So, I have a concept of the object, which picks out what I’m referring to, and thus what I can be wrong about. But since this is my concept, it must also be subject to the possibility of error.

4. But according to (2), consciousness cannot be consciousness if its object is for it, and so it must suppress this fact. So, in order to allow that my belief might misrepresent its object, I must ignore the possibility that the concepts that pick out this object might themselves misrepresent it.

What does this mean? Say, for example, that you and I are having a disagreement about whether or not truth is beauty, then we need to be able to agree about what we mean by “truth” and “beauty” in order to have the disagreement. There’s got to be enough of a common understanding between us to pick out that which either of us might be wrong about. It is a condition of the possibility of error that there are some things that we aren’t opening up to the possibility of error, a shared understanding of the meaning of our terms that prevents us talking past one another. But this shared understanding looks like it’s been stipulated. If we’ve simply stipulated what black holes are, how can we be engaged in an inquiry in which the universe might reveal that our theories about black holes are mistaken, rather than simply having set up a fiction? Why is it any different from talking about unicorns? Unicorns only have horns because we’ve somehow stipulated that they do. Another way to put this is that unicorns don’t really have horns, because they aren’t real.

AM: How does your picture of the social nature of concepts fit in with something like Ian Hacking’s distinction between natural kinds and social kinds?

PW: Often, the way in which people want to talk about this difference between natural and social kinds – things like on the one hand like uranium and radioactivity, and on the other hand, say, gender and money – is to find a metaphysical distinction between them. They want to say that there is some difference in the way these things really are, that there is a real difference between what is real and what is unreal – as if you could look at the properties of these things and then develop a second-order classification of the difference between them on the basis of those properties. I have a different way of looking at this: rather than trying to look at what we take to be the things that the concepts are picking out, I think we should look at how the concepts are picking them out. We should try to differentiate the ways in which concepts strive for objectivity, and then we can classify them on the basis of the extent to which they achieve it. Reality isn’t a property of things, but an ideal aimed at by some concepts but not others.

A lot of people want to argue that race is not a real biological category, but that this doesn’t mean that it’s a fiction.

To take an interesting example, a lot of people want to argue that race is not a real biological category, but that this doesn’t mean that it’s a fiction. Part of the awkwardness here is that the very term “reality” has a bunch of different connotations and it can be used in a variety of ways. You can roughly distinguish these by the types of “unreality” they’re opposed to. There’s a difference between talking about what is real as opposed to fictional and what is real as opposed to fake, and a difference between what is real as opposed to illusory and what is real as opposed to what is insubstantial (or epiphenomenal). Fictions are absolutely unreal, because non-existent. Whereas fakes are relatively unreal, because insufficiently similar to a standard, such as a real designer handbag. The distinction between illusions and epiphenomena is similar. Illusions are mere appearance, such as a stick that appears bent in water but definitively is not, whereas epiphenomena are secondary effects of some deeper process that might not immediately be apparent, making them in some sense less real – such as a rainbow produced by refraction in water vapour.

When it comes to race, the rough consensus is that it doesn’t refer to some substantial biological essence, but is some sort of appearance. It might not exactly be an illusion, but it doesn’t have the explanatory power of categories like species, sex, or blood type. Put differently, even if you can consistently label people by race, the classification doesn’t have many useful implications. Nevertheless, there are many who want to maintain some distinction between the real and the fake here. They want to retain some notion of the reality of race in the sense of authenticity, even if they want to reject the reality of race in the sense of essence.

AM: The distinction between authenticity and essence is interesting as philosophers have tended to think that’s it is essence or nothing, so if you throw out essence, you lack any possibility of achieving some kind of objectivity.

PW: Again, my view is that we see this distinction between authenticity and essence as something that emerges out of the structure of the practices through which we aim at objectivity, rather than something that explains them. That might not sound too precise, but let me try and elaborate. One way of distinguishing natural from social kinds is by appealing to a notion of mind dependence. Take money: if there weren’t humans around to treat these pieces of paper this particular way then there wouldn’t be any money. But the same might be said of economies, business cycles, depressions, and a bunch of phenomena whose existence is dependent on human commerce, but whose nature we have little or no control over. We might stipulate the properties of money in such a way that we can distinguish real from counterfeit, but the processes that it becomes enmeshed in move in ways we can’t simply stipulate. So, a better way of differentiating natural and social kinds is in terms of attitude dependence: it’s not whether the existence of the kind is dependent on the existence of human minds that’s at issue, but whether the way the kind is, or what is true about it, is dependent on the way we take it to be. A five pound note is worth five pounds because we take it to be worth that much.

AM: So if that’s not a metaphysical distinction, what kind of distinction is it?

PW: An epistemological one! For me, truth is the fundamental concept underpinning all these different ways in which we talk about reality. As I hinted earlier, the key thing for understanding truth is error, as there is no possibility of truth without the possibility of error. So the way to come at thinking about things that we don’t want to classify as “fully real” in the sense of having an underlying essence that is independent of the social domain is to explain the differences between the ways we allow ourselves to be wrong about them. In short, the kinds of error that we’re capable of making about uranium are fundamentally different from the kinds of error we are capable of making about money. The trick is not to take for granted that we already understand the nature of objectivity, because that suggests that its absence is what needs explaining. From my perspective, the difficulty isn’t explaining how social kinds fall short of natural ones, but explaining how natural ones exceed social ones. How is it that we’re capable of being more wrong about black holes and species than money and race?

The kinds of error that we’re capable of making about uranium are fundamentally different from the kinds of error we are capable of making about money.

AM: So does your idea of semantic deference have any role when thinking of these social categories?

PW: It’s all about the way in which discursive practice is structured in terms of authority and responsibility. We all have authority over which claims we take to be true, and when we assert them we authorise others to take them to be true too. But, as we saw above, in order to make a truth claim we need to allow for the possibility that it is false, which means relinquishing the authority to stipulate that what we say is true. We do this by undertaking a corresponding responsibility to justify our claims in response to challenges. This takes roughly two forms: inferring the claim in question from other claims that have yet to be challenged (e.g., “The banana is ripe because it is yellow”), and deferring to somebody’s authority (e.g., “My father is sure the banana is ripe”). The authority that is deferred to can take various forms, including the authority to stipulate that certain claims are true by definition (e.g., “This is just what we mean by ‘ripe’ in this house”). This can as easily be communal as individual, such as when we appeal to common knowledge or social mores. The crucial thing is that deference involves appealing to claims about people’s attitudes in a way that seems to make truth dependent upon them.

Fiction is the easiest case to begin with. If we get into a debate about obscure details of the Harry Potter universe, then we are inevitably going to end up appealing to claims about JK Rowling’s attitudes: what she intended to be the case. Of course, there are complexities involved in such interpretation. For example, Rowling might have certain contradictory ideas that invite us to interpret things in a slightly different or nuanced way, letting us disagree with her in limited ways. But there’s a fundamental sense in which all this remains dependent upon claims about attitudes, about what she takes to be the case. This dependence can become more diffuse, such as debates about mythology, like the nature of the Norse gods. This is still fiction, but we’re not necessarily going to be tracing back all of these debates to ultimate claims about specific individuals’ attitudes about Thor. We might draw a similar comparison between laws explicitly instituted by legislators and interpreted by judicial institutions, which can create and regulate social kinds like currencies, and the norms implicitly maintained in a decentralised manner without specialised experts, which can sustain and authenticate social kinds like classes. What distinguishes the type of truth at issue in all these cases is that, although they’re open to the possibility that I might be wrong, they’re not open to the possibility that everyone might be wrong. Objectivity demands that we go further.

AM: How do we get there, then?

PW: The first step in this direction is showing that it’s possible to eliminate appeals to claims about attitudes from at least some forms of deference to epistemic expertise. Sellars shows us how to do this for observational testimony. For example, if we’re speculating about the weather, I can say “I think it’s raining because Steve’s outside and he’s called me to say that it’s raining”. Technically, I’m deferring to his opinion here. But we can go a step deeper and we can start thinking about Steve as a causal system that responds differentially to the weather in certain reliable ways, analysing his expertise and reducing deference to inference. In principle, I could even explain why Steve can reliably tell that it’s raining in a way that doesn’t appeal to his or anybody else’s attitudes, but merely to further observations, digging into the neurology of his perceptual mechanisms and looking for reasons why when he experiences certain sensations he produces certain words. This also means that Steve’s testimonial authority is open to challenges based on how his perceptual mechanisms are functioning. If you respond that “Steve’s taken a lot of LSD”, that’s a good reason to discount his testimony. In essence, we can look at Steve the same way we look at an electron microscope. We can treat him as an instrument, and argue for or against believing his readings in the same way. By the same logic, we can treat ourselves as instruments, accepting that our own observations are justified and subject to challenge in precisely the same way as Steve’s.

The second step is more difficult, and it amounts to solving the problem posed by Hegel’s account of natural consciousness: how do we eliminate the residual attitude dependence implied by deference to semantic expertise? There are two basic insights here. On the one hand, we must recognise that fixity of meaning is relative to the topic under discussion. Although we have to suppress the difference between our concept and what it represents in order to use it in representation, we can temporarily suspend this suppression. We can criticise the adequacy of definitions, the consistency of techniques of classification, or the coherence of the socio-cultural infrastructure that ties these together. To return to the earlier example, we might pause our debate about whether truth is beauty in order to get clear on what we mean by “beauty” and whether there even is a common thread uniting the various different senses in which that word gets used. We can bring the contents of our concepts into question, in order to revise them – we just can’t bring them all into question at the same time. We always have to use some to argue about the adequacy of others.

We can bring the contents of our concepts into question, in order to revise them – we just can’t bring them all into question at the same time.

On the other hand, we need to see that what is characteristic of empirical concepts is that the world can force us to revise them. Concepts encode patterns of inference, but these inferences can be contradicted by experimental evidence. Consider the concept of “particle” which was treated as mutually incompatible with “wave” until experimental evidence showed entities treated as particles behaving like waves and vice versa. The history of science contains a plethora of such shifts, where not only which things fall under a concept can change but also what it means for them to do so. What we have here is a game that is set up in such a way as to require us to change the rules we’re playing by, creating a kind of openness to the world through which it can tell us that we’re all playing the game wrong.

This gives us some sense in which the evolving meaning of natural kind concepts is tracking some essence that exceeds them, without implying that we have any positive access to that essence, or thereby any certainty that there is something we’re tracking. Some conceptual trajectories lead to dead ends, like “phlogiston”, “aether”, and “hysteria”. By contrast, in the case of social kind concepts like “(racially) black” or “(legally) woman”, it’s not to say that we can’t revise them or that there aren’t good reasons to revise them, but that the reasons for doing so are fundamentally different.

AM: And what are they? Take a concept like “woman”. This seems to sit somewhere between the two because on the one hand it seems like a clear biological category, and on the other hand it goes well beyond biology. What role should the philosopher be playing in helping to clarify what is going on in debates about such concepts?

PW: Well this is the question, right? The reasons that most thinkers are giving for changing these are social ones – they are about how the concepts affect the ways in which we live and treat each other. I’ve thus far focused on the role that concepts play in observation and inference, as this is what’s central to empirical concepts, but concepts can also play a role in action. In other words, some concepts have practical implications. This is more obvious with abstract value concepts such as “justice” and “beauty”. For example, if I say that one painting is more beautiful than another, this implies that, all else being equal, if you were in a burning gallery and could only save one of the two, you should save the more beautiful one. Value is what motivates action. But there are a lot of more concrete concepts with normative content. One can’t grasp the concept “table” without understanding how one ought to use it, and one can’t grasp the concept “cashier” without understanding how one ought to interact with them. To tinker with these concepts is to actively modify the furniture of the social world.

The recent controversy over the concept “woman” is a good example of how the empirical and normative dimensions can be intertwined, insofar as we have a biological classification that is supposed to have social implications. I don’t want to rehearse the details of this debate, because I’m hardly an expert. But I think I can offer two useful philosophical points, based on what I’ve said so far. First, the biological concept of “sex” has undergone precisely the sort of changes characteristic of the pursuit of objectivity, revealing complexities that cut against the intuitions encoded in earlier practices of classification. Insofar as there is anything like an essence here, there is nothing “common sense” about it. Second, the naturalistic fallacy still applies: one cannot directly infer an ought from an is. The normative content of “woman” cannot be derived from its empirical content. Even if the latter involves some conception of biological function, for example that the role of the female is to produce larger gametes (eggs), this function is purely explanatory. It’s the same as saying that the function of a heart is to beat. It tells us nothing about how we should act or the kinds of practical attitudes we should adopt towards those classified as women. This leaves us free to modify either the practical implications or the classificatory scope of the concept as we choose.

Concepts don’t exist in vacuums. They’re fundamentally determined by their relations to one another, and it’s hard to tinker with them in isolation.

AM: All this seems to suggest that we have a lot of control over how we revise these kinds of concepts.

PW: There are some who suggest that we have more or less infinite stipulative control over our concepts. This may not do his views justice, but I’ve had discussions about this sort of thing with the philosopher Liam Bright who comes from a Carnapian perspective, and his answer to a lot of questions about problematic features of our concepts is something like, “Well, we could just not.” For example, afropessimists maintain that the concept of “human” has historically been defined in terms of what it excludes, and that “blackness” forms its boundary, such that the categories through which the rest of Western society identifies itself are premised upon systemic subjection of those categorised as black. Bright’s response is that, even if this were true, we could simply choose to use “human” in a way that is non-oppositional. I don’t exactly agree with the afropessimist analysis, and I’m sympathetic to Bright’s conceptual voluntarism. However, I also have some reservations about just how easy it is to effect these sorts of change. On the one hand, concepts don’t exist in vacuums. They’re fundamentally determined by their relations to one another, and it’s hard to tinker with them in isolation. On the other, if you take seriously that they’re social practices, then the difficulty is propagating changes in the way they’re used through the community. This is easier for technical terms whose networks of semantic deference are institutionalised, but much harder for common terms in the wild. The task faced by would be “conceptual engineers” is quite a daunting one.

Peter Wolfendale is an independent scholar whose work focuses mainly upon the intersection between the methodology of metaphysics and the structure of rationality, but also includes foundational topics in the philosophy of value, ethics, aesthetics, computer science, and social theory. Website: Twitter: @deontologistics

Anthony Morgan is editor of The Philosopher.


From The Philosopher, vol. 109, no. 3 ("Concept and Reality").

If you enjoyed reading this, please consider becoming a patron or making a small donation.

We are unfunded and your support is greatly appreciated.


bottom of page