The Janusian Nature of Moral Ought

Abstract

Moral contextualism about “ought”-sentences is a semantic thesis that takes the content of moral “ought” to be a function of contextually relevant parameters. I aim to provide a theory of ought-judgments at the level of thought that supplements the contextualist understanding of moral “ought”-sentences. To this aim, I suggest understanding the concept of ought as an indexical concept for which a phenomenological profile plays an extension-fixing role. Then, I will argue how my suggested view can provide a theory of ought-judgments that is in harmony with moral contextualism.

An increasingly popular approach to the semantics of moral “ought”-sentences is moral indexical contextualism, i.e., the view that takes the semantic content of moral “ought” to be a function of contextually relevant parameters. Despite recent advancements in refining and defending their account,¹ contextualists have not yet developed a sufficiently detailed view of the counterparts of “ought”-sentences at the level of thought. My aim in this paper is to provide a theory of moral judgments at the level of thought that supplements the contextualist understanding of moral “ought”-sentences.

This task is more urgent for the contextualist project than it might initially seem. Some pivotal concepts and distinctions suggested by the contextualists to explain “ought”-sentences can only make full sense by being in harmony with a defensible theory of ought-judgments—the judgments at the level of thought that use ought² (i.e., the concept of ought). In particular, the role of moral motivation and its relation to ought-judgments remain unexplained in the contextualist framework. Thus, in the absence of such a theory, contextualism suffers from a lacuna. Or so I will argue. In this paper, I aim for three separate goals. First, I will offer a theory of ought-judgments that is in harmony with the contextualist account of moral “ought”-sentences. Second, I will offer a non-expressivist account of moral judgments that is compatible with motivational internalism. Third, I will account for the irreducibility of ought to physical or natural concepts.

My suggested theory is that ought should be understood as a Janusian concept,³ i.e., an indexical concept for which a phenomenological profile plays an extension-fixing role. By taking this approach, we endorse what David Chalmers calls conceptual functionalism regarding ought. Conceptual functionalism is true of a concept if it picks out its extension in virtue of its causal role with respect to certain phenomenal experience.

This understanding of the concept of ought enables me to develop a theory of ought-judgments matching the contextualist account of “ought”-sentences. The paper is in two parts. In the first part, I put forward a version of contextualism and investigate its implications for a theory of ought-judgments. The discussion in this part provides us with a clear picture of what we should expect from a theory of ought-judgments that is in harmony with our adopted version of contextualism. In the second part, I develop the Janusian take on ought-judgments and defend it against some objections.

1 What is Moral Contextualism?

Moral contextualism about “ought”-sentences, in a nutshell, is the idea that the semantic content of moral “ought” is relativized to certain contextually relevant parameters. Indeed, contextualism can be used to account for different readings or flavors of “ought”, e.g., legal “ought” or prudential “ought”. But for the sake of simplicity, I only focus on moral “ought” in this paper. Thus, whenever I mention “ought”-sentences (or later ought-judgments), the reader can assume that I am making a point about the moral reading of them, unless I explicitly mention another reading.

In the literature, there are different models of context-sensitivity based on which we can understand the alleged context-sensitivity of “ought”. The most common approach among “ought”-contextualists is to take “ought” to be indexical à la Kaplan (1989).⁴ This is to take “ought” to have a context-invariant component, namely character, determining its content in any given context. The content of “ought” varies from context to context, though its character remains the same (Dreier, 1990, 2009; Worsnip, 2019). This is why this version of contextualism sometimes goes under the name of indexical contextualism.

There can be different takes on what the contextually relevant parameters are. The usual candidates are the salient value system and the relevant body of information in the context. For contextualists, the first parameter always plays a role in determining the content of “ought”- sentences. By contrast, the second parameter is only relevant when an “ought”-sentence reflects the subjective sense of moral ought. The subjective sense of “ought” concerns what an agent ought to do given the information available to her. According to contextualism, the semantic content of “ought”, reflecting the subjective sense of “ought”, is information-sensitive—its content can vary according to the variation in the relevant body of information. However, as Dowell points out, contextualists can also accept an information-insensitive use of “ought” that reflects the objective sense of moral ought (Dowell, 2013: 158). This sense of ought is about what an agent ought to do given all relevant facts. The content of information-insensitive moral “ought” is merely a function of a contextually relevant value system (henceforth, CVS), not any information-related contextual parameters, e.g., the available body of information.⁵ In this paper, our discussion is focused on judgments regarding the objective moral ought. Thus, the CVS is the only relevant contextual parameter for our discussion.

Interests in contextualism was fueled by two main considerations. First, it matched perfectly with a prominent analysis of “ought” offered by Kratzer (1977, 1981). This is a plus for the view since Kratzer provides a model that systematically explains the different uses of “ought”, including epistemic “ought”.⁶ Second, and perhaps even more importantly in its early phase of development, contextualism could accommodate motivational internalism, i.e., the view that sincerely asserted “ought”-sentences are intrinsically correlated with the speaker’s motivation to act accordingly (Dreier, 1990:7).⁷

The link between motivational internalism and contextualism is directly related to an important question: what is it that makes a particular value system the CVS in a given context? Perhaps the most straightforward answer to this question is to take the value system endorsed by the speaker to be the CVS. This version of contextualism, often called speaker contextualism, can easily accommodate motivational internalism. Consider the following judgment:

(1) One ought not to torture animals

According to speaker contextualism, the content of “ought” is a function of the value system endorsed by the speaker. Thus, (1) is true if and only if it is implied by the value system endorsed by the speaker. Assuming that (1) is true given the speaker’s value system, she also has a motivation to act accordingly. After all, she has endorsed the value system.

In recent years, however, contextualists have become divided on the issue of motivational internalism.⁸ A reason for the decreasing popularity of motivational internalism among contextualists is the fact that we can use “ought”-sentences to report judgments associated with value systems that we reject. To see this, imagine Jim and Jordan, two militant atheists, having the following conversation about their friend, Ali, a Muslim who believes in the divine command theory of morality:

(2) Jordan: Why didn’t Ali come to lunch with us today?

Jim: Ali ought to fast in Ramadan. Didn’t you know that?

Here, taking “ought” as used in (2) to be a function of the speaker’s value system gives us wrong results. If the extension of “ought” in (2) were determined by the speaker’s value system, the first sentence in (2) would be clearly false. To avoid this problem, we should maintain that “ought” is not relativized to the speaker’s value system, but to that of Ali.

Given this and similar observations, some contextualists have renounced the idea that “ought”-sentences are always relativized to the speaker’s value system. Instead, they take the CVS to be the value system endorsed by the salient individual (who might not be the speaker) or the salient group in the context. In the above example, Ali, not the speaker, is the salient individual and the content of “ought” is relativized to his value system.

But these alternative accounts of what it takes to be the CVS in a given context are incompatible with motivational internalism. The reason is simple: these alternative accounts allow the speaker to sincerely and assertorically utter “ought”-sentences without being motivated to act accordingly (Worsnip, 2019: f38). This is why in recent years a growing number of contextualists have distanced themselves from the position most notably defended by Dreier, i.e., that “the main motivation” for contextualism is its ability to incorporate motivational internalism (Dreier, 2009:80). This move, however, deprives contextualism of one of its main advantages, i.e., accommodating motivational internalism. Thus, we need to investigate whether there is a middle path that can account for the above observation without decoupling contextualism from motivational internalism. We will return to this issue in the next section.

2 The Implications of Contextualism for Ought-Judgments

The main aim of this paper is to develop a theory of ought-judgments that is in harmony with moral contextualism. To this end, we need to work out the implications of accepting moral contextualism for our theory of ought-judgments. By investigating these implications in this section, we will pave the way for our discussions in the later sections in which we will develop our suggested theory of ought-judgments.

The first implication to consider regards the indexicality of ought. Does the contextualist need to take ought to be an indexical concept? I believe the answer is positive. Contextualism about “ought”-sentences does not square with taking ought to be a single (or a set of) context- insensitive concept(s). Though some might find this claim to be intuitive, there are skeptics who find it in need of argument. My argument for this claim goes as follows. Given the contextualist account of “ought”, contextualists have three main alternative accounts of ought. First, they can embrace an approach according to which ought is a single concept with a single extension. Second, they can claim that there are various distinct context-insensitive moral oughts, i.e., we have different concepts of moral ought, e.g., the Kantian ought, the utilitarian ought, etc. Third, they can accept that moral ought is a single but indexical concept. I will argue that the first and second theses do not sit well with contextualism.

Consider the following examples:

• (1) “One ought not to lie.” (Uttered in a context in which Kantianism is the CVS)

• (2) “One ought not to lie.” (Uttered in a context in which utilitarianism is the CVS)

By sincerely asserting each of these judgments, the speaker expresses his corresponding judgment at the level of thought. According to contextualism, the content of (3) and (4) are not the same. Consequently, the content of ought-judgments expressed by these sentences cannot be the same either. However, by taking ought to be a single invariant concept, we face problems in accounting for the difference between these two ought-judgments. If all concepts used in both ought-judgments expressed by (3) and (4) are the same and their extensions are also identical, their content must be the same too. Therefore, no contextualist can accept that we have a single context-insensitive ought.

The second alternative is more successful in explaining why the contents of the thoughts expressed by (3) and (4) differ. On this option, one can claim that different oughts are deployed in the thoughts expressed by (3) and (4). This take, however, is at odds with the contextualists’ criticism of one of their main rivals, namely the ambiguity theory—the view that takes “ought” to be ambiguous between different meanings. I believe the reasons provided by contextualists against the ambiguity theory as a semantic thesis about “ought” can easily be translated to the corresponding arguments against the second alternative which is a thesis about ought. To show this, I briefly discuss these reasons.

Contextualists give us two reasons for why taking “ought” to be indexical is a more defensible view than the ambiguity theory. First, there is a noticeable systematicity behind the semantic contributions of “ought” to a sentence containing it in different contexts. The notion of character can very well explain this systematicity. But the proponents of the ambiguity theory lack an equally plausible account of this systematicity (Worsnip, 2019:3102). Second, taking “ought” to be ambiguous makes learning it unnecessarily difficult. Assuming that “ought” is an indexical, a language learner merely needs to grasp its character. But if it was an ambiguous term, she would need to learn the numerous meanings of “ought” one by one (Chrisman, 2015:42). Though both reasons mainly target the ambiguity theory about “ought”, they also show why the view that there are different oughts is untenable.

First, there seems to be a systematic connection among tokens of ought used in the judgments of individuals endorsing different value systems, say, Kantianism, act utilitarianism, etc. We can see a connection among the roles played by the concepts of ought in determining the content of ought-judgments. Therefore, this connection goes beyond the fact that we express all these concepts with the same term in English. By contrast, bank (referring to a financial institution) and bank (the slide of a river) have nothing to do with each other save the fact that the same word is used to express them in English. By taking ought to be an indexical concept, we can explain this systematic connection. This would not be possible if we posited multiple invariant oughts. Second, accepting the idea that there are multiple oughts makes the practice of grasping concepts unnecessarily difficult. If contextualists take issue with the ambiguity theory because it makes the practice of learning a language unnecessarily difficult, they should appreciate the similar concerns when it comes to the practice of grasping concepts. Thus, the contextualist also needs to reject the view that there are multiple invariant oughts.

Given our discussion, the first and the second alternatives should be abandoned for the third. Therefore, contextualism about “ought”-sentences sits most naturally with the view that ought is an indexical concept. This is the most significant implication of accepting contextualism for our theory of ought-judgments.

Another discussion that is directly relevant to our theory of ought-judgments is the internalist/externalist debate within the camp of contextualists. In the first section, I maintained that the most significant problem for the internalist-leaning contextualists is our ability to use “ought”-sentences to report on value systems that we reject. In such cases, we will not be motivated to act according to our judgments. This is an observation that I find hard to resist. Perhaps, the simplest way to respond to this problem is to join the externalist-leaning contextualists and reject motivation internalism.

However, this move has a significant cost, i.e., contextualism loses one of its important advantages over its rivals. Contextualists initially put forward their view as a theory that has the advantages of both expressivism and descriptivism without suffering from their problems. One of these alleged advantages is the ability of contextualism to incorporate motivational internalism. Thus, it would be premature to decouple contextualism from motivational internalism so quickly. We should ask whether there is a chance of finding a way to reconcile motivational internalism with the observation that one can report on value systems that one rejects.

A way out of this problem is to limit the scope of motivational internalism to a subgroup of “ought”-sentences.⁹ An oft-made distinction between the normative and descriptive uses of “ought” can be helpful here. One can use “ought” to report moral judgments to which one is not committed. This is called the “reportive” or “descriptive” use of “ought”. For instance, a Kantian can use “ought” to report a utilitarians’ moral convictions. By contrast, “[n]ormative uses present the speaker as endorsing the norms that justify” her claim (Silk, 2017:227).¹⁰ Given this distinction, we can take motivational internalism to be true of the normative use of “ought”-sentences, not their descriptive use.

For the rest of the paper, I will assume a version of contextualism that posits two uses of “ought”-sentences: descriptive and normative. In each use, “ought” has a different character and its content is sensitive to different contextual elements. In its descriptive use, the content of “ought” is sensitive to the value system endorsed by the salient individual in the context. By contrast, in its normative use, the content of “ought” is a function of the value system endorsed by the speaker. I take motivational internalism to be true of the normative use of “ought”- sentences.¹¹

Making the above distinction enables us to account for our ability to make judgments based on value systems that we reject. This ability, however, is not limited to our judgments at the level of language. We have a similar ability when it comes to making ought-judgments. This gives us a good reason to recognize a corresponding distinction at the level of thought. On the one hand, a subject can make ought-judgments that reflect the value system endorsed by her. On the other hand, she can make ought-judgments describing a value system rejected by her.

It is important to note that one cannot let both types of ought-judgments enter one’s deliberation process. The reason is simple. Given the contextualist framework, one can potentially make ought-judgments associated with numerous value systems. Taking all these ought-judgments into account while making a decision brings the process of deliberation to an impasse.¹² Hence, the practice of making moral judgments loses its key function: to provide guidance on what to do.¹³

Moreover, it is absurd for the agent to take into account the second type of ought-judgments in her decision-making. After all, these are the judgments associated with value systems that she does not endorse. Therefore, only the first type of ought-judgment has action-guiding authority and can enter the process of deliberation. This is why I call the first type authoritative judgments and the second type practically inert judgments. Given the above discussion, it is safe to assume that a normative use of an “ought”-sentence, on the condition of being sincere, expresses an authoritative judgment. By contrast, the descriptive use of an “ought”-sentence expresses a practically inert judgment.

Here, a question arises as to whether authoritative ought-judgments are intrinsically motivational. At least prima facie, there are good reasons to believe so. By taking authoritative ought-judgments to be intrinsically motivational, we can kill two birds with a single stone. First, we can explain why these judgments are practically relevant and enter the process of deliberation. Second, this idea is in harmony with accepting motivational internalism regarding the normative use of “ought”-sentences. After all, “ought”-judgments used normatively express authoritative ought-judgments. If there is a necessary connection between authoritative judgments and the moral motivation to act accordingly, it is no surprise that the normative use of “ought”-sentences is correlated with having the motivation to act accordingly. But we cannot simply assume that this is a feature of authoritative ought-judgments. A satisfactory theory of ought-judgments needs to provide an account of why authoritative ought-judgments are intrinsically motivational.

Now, we can list the three implications of accepting contextualism for our theory of ought- judgments:

• I. Recognizing the indexicality of ought and ought-judgments.

• II. Positing a distinction between authoritative and practically inert ought-judgments.

• III. Providing an account of the necessary connection between authoritative ought-judgments and moral motivation.

The next three sections develop a view of ought-judgments that has these three features.

3 Janusian Concepts

In this section, my goal is to discuss Janusian concepts, i.e., indexical concepts for which a phenomenological profile plays an extension-fixing role. My primary goal is to show that the model provided by Janusian concepts explains the moral ought, as well as ought-judgments. However, the model also applies to a range of other concepts including a certain type of taste concepts, e.g., salty. In fact, the examples of taste concepts provide us with a more intuitive grip on the nature of Janusian concepts. I rely on examples of taste concepts to elaborate on different aspects of Janusian concepts. Finally, I will briefly explore the links and differences between the Janusian framework and a prominent model of concepts in the literature, namely the response-dependence approach.

A way to conceptualize a property is to understand it in terms of its causal impact on forming a particular experience. If you take a concept to pick out its property in this way, you are conceptual functionalist about that concept (Chalmers, 2012:322). However, not all such concepts are indexicals. For instance, consider C, a concept whose extension is the property that induces the taste of saltiness in Wittgenstein in 1936 in the actual world. Assuming (contrary to Lewis) that “actual” is not an indexical term, such a concept is not indexical. But conceptual functionalism is true of it. In the rest of this section, I will discuss a range of indexical concepts about which conceptual functionalism is true. For reasons to be made clear shortly, I call these concepts “Janusian”.¹⁴

Our starting point is a familiar distinction between phenomenal concepts, e.g., pain, sweetness, and redness, and physical concepts such as salt, sugar, and water. There is a sizable literature on the differences between these two types of concepts.¹⁵ Here, I only focus on two theses:

• (a) Irreducibility Thesis: a phenomenal concept is not analyzable in terms of physical concepts. Thus, in principle, there is no a priori path from a phenomenal concept to any physical concept.¹⁶

• (b) Semantic Stability: phenomenal concepts, contrary to physical concepts, are semantically stable—their extension is the same regardless of how the facts are or might be.

We can briefly elaborate on each thesis. The first thesis maintains that phenomenal concepts are irreducible to physical concepts. This thesis is supported by the so-called knowledge argument famously articulated by Frank Jackson (Jackson, 1982). The argument is based on a thought- experiment in which a color scientist leaves her black and white room for the first time after learning all scientific facts about colors. The basic idea is that she acquires a new piece of knowledge in her first encounter with colored objects, despite knowing all relevant scientific facts about them beforehand. The argument originally aims for a more ambitious goal, namely to establish that phenomenal properties are irreducible to physical properties. Though its success in achieving its original aim is a matter of controversy, many believe that it establishes a weaker thesis: phenomenal concepts are irreducible to physical concepts.

The second thesis holds that the extension of a phenomenal concept such as pain is a constant function of the subject’s environment. By contrast, the external environment plays a role in fixing the extension of physical concepts like water. The extension of a physical concept is, therefore, “actual‐context dependent” while the extension of a phenomenal concept “can be determined independently of any empirical discoveries” (Balog, 2009:299).¹⁷

There is, however, a type of indexical concept that can be categorized as a third type. These are indexical concepts for which a phenomenological profile plays an extension-fixing role. Concepts like salty (i.e., things that induce the taste of saltiness) and pain-inducing can be categorized in this third category. As I will argue shortly, on the one hand, like phenomenal concepts, these concepts are irreducible to physical concepts. On the other hand, they are not semantically stable, i.e., their extension is not a constant function of the subject’s environment. In this way, the first thesis mentioned above is true of this type of concept, but not the second. Due to this in-between nature, I call them Janusian concepts.

Before arguing for each of these claims, I focus on illustrating the notion of Janusian concepts with some examples. Consider the following concepts: saltiness, salted, and salty. saltiness refers to a phenomenological profile (similar to pain or sweetness), while salted stands for what contains sodium chloride.¹⁸salty is a concept whose extension is determined by what induces saltiness experiences in the relevant subject. Thus, salty relativized to an average human being is co-extensional with salted. This concept can have different extensions relative to different subjects. We can even imagine creatures who find sugar to be salty.

Taking salty to be an indexical concept enables us to account for the variation of its extension. We can hold that what determines the content of salty is its character, i.e., what induces the taste of saltiness for the contextually relevant subject at the contextually salient time. The extension of salty can vary if the contextually salient subject changes. But even for a single subject, the extension of salty can change in the course of her life. For instance, due to Covid, I could lose my sense of taste, and salted things could cease to taste salty for me. Therefore, the concept of salty is relativized to both the contextually salient time and subject.

Similar to salty, I believe ought is a Janusian concept. To take this concept to be Janusian, we need to find a phenomenological profile that plays an extension-fixing role for it. An option for the relevant phenomenological profile is the feeling associated with moral motivation. My assumption is that similar to pain or saltiness, moral motivation has a distinct feeling to it enabling the subject to distinguish it from other phenomenological profiles. This phenomenological profile plays an extension-fixing role for ought.

Here, an immediate objection comes to mind: unlike saltiness or pain, moral motivation lacks a distinct phenomenology. In fact, the idea of characterizing moral emotions via their phenomenology is a long-debunked Ayerian view that nobody buys today. Thus, characterizing the extension-fixing mechanism of moral ought based on it is, at best, dubious. A full treatment of this objection goes beyond the scope of this paper. However, I can offer the outline of a response which, at least, licenses some degree of optimism regarding my view.

Often, we identify the phenomenology of a mental state as the feeling or internal experience associated with that very specific mental state in isolation from other internal experiences. However, we can embrace a more holistic picture of phenomenal states that identifies them partially in terms of their relations with other phenomenal states.¹⁹ For instance, a feeling of grief that comes with a sense of pride feels different from a feeling of grief without it. Indeed, sometimes, one needs to reflect on one’s phenomenal states to detect the presence of other phenomenal states and appreciate their impact on each other. This is why sometimes a session with a therapist can have such an illuminating effect on the understanding of our feelings.

In the case of moral motivation, we feel a motivation to act that comes with a sense of prospective guilt. The sense of prospective guilt is what the subject feels when she is aware of the fact that defying a motivation to take a course of action will lead to a full-blown sense of guilt. The phenomenology of the sense of prospective guilt is different from that of guilt, and its presence makes our motivation to take a course of action feel different. What makes moral motivation truly distinctive is its correlation with this sense of prospective guilt. Such a feeling is sometimes immediately accessible, and sometimes can be found upon reflection.²⁰ However, my claim is that if an agent is merely motivated to take a course of action and cannot find any feeling akin to the sense of prospective guilt in herself, her motivation cannot be classified as moral motivation.

So far, I have explained the phenomenological profile that plays an extension-fixing role for ought without addressing the question of what its character is. For now, I bracket this question, to return to it in the sixth section. Instead, I will focus on a different question, i.e., do Janusian concepts constitute a third category that cannot be reduced to phenomenal or physical concepts? To show this, it is enough to argue that the irreducibility thesis is true of Janusian concepts, but not the semantic stability thesis. The best reason to think that Janusian concepts are irreducible to physical concepts is the fact that one can run the knowledge argument for them. Imagine a taste scientist who has studied all scientific facts about the human experience with the taste of sweetness and also every scientific fact about honey. However, she has never tried anything sweet in her life. After learning all the relevant scientific facts, she puts some honey in her mouth. In this way, she comes to know something about honey that she did not know before: it induces this very feeling of sweetness that she can detect in her phenomenal experience right now. In other words, she learns that honey is sweet. In this way, she acquires two concepts during this experience: a phenomenal concept, namely, sweetness, and a Janusian concept, i.e., sweet. These are the concepts that she could not obtain by merely studying scientific facts about taste. This line of argument closely mimics Jackson’s argument for phenomenal concepts. In both cases, the agent acquires a new concept that cannot be obtained by deploying physical concepts.

The semantic stability thesis is not true of Janusian concepts, however. To see this, imagine two agents whose epistemic situations are qualitatively identical. The first one lives on the earth and finds only water to be salty, while the second one, who inhabits twin earth, finds only t-water to be salty. Here, the extension of salty is not the same for these two individuals since different substances induce the taste of saltiness for them. Thus, salty is not semantically stable. Consequently, one thesis is true of Janusian concepts, but not the other. Ipso facto, Janusian concepts constitute a third category of concepts.

Here, we can see that taking ought to be a Janusian concept comes with two important theoretical advantages. First, I claimed that the indexicality of “ought”-sentences lends support to taking corresponding judgments at the level of thought to be indexical too. Being a Janusian concept, ought is indexical and so are the judgments in which this concept is deployed. This brings us a step closer to our main target: providing a theory of ought-judgments that is in harmony with contextualism.

Second and more importantly, it satisfies the widely held intuition that ought is irreducible to physical concepts. This intuition is shared by many and is taken to be one of the main advantages of the expressivist account of moral thought.²¹ Expressivists are proud of the fact that their view accounts for the irreducibility of moral concepts like ought to physical concepts without giving up on naturalism. However, taking ought to be a Janusian concept provides us with an alternative option with even more advantages. Given the irreducibility thesis, not only can our theory satisfy the aforementioned intuition but also, to some extent, it can demystify it. There is nothing mysterious about ought that makes it irreducible to physical concepts. The reason is that a phenomenological profile is used in its character. In this regard, it is similar to other Janusian concepts like salty. At the same time, we do not need to give up on naturalism since the irreducibility thesis and Jackson-style knowledge arguments merely show the irreducibility of certain concepts, not that their extension is non-physical. Thus, by adopting my suggested theory, the contextualist can account for the irreducibility of ought while explaining its indexical nature.

One should avoid an important misunderstanding. The Janusian account does not offer a metaphysical characterization of the extension of salty or ought. It merely offers a semantic account of how this concept latches onto its extension. Thus, one should not confuse this theory with so-called dispositional views that define a property based on human responses to it.²²

Moreover, one should not confuse Janusian concepts with so-called relational phenomenal concepts. One can form a concept of a phenomenal property by referring to the objects inducing this property. For instance, Chalmers gives the following characterization of red as a relational phenomenal concept: “the phenomenal quality typically caused in me by paradigmatic red things” (Chalmers, 2004:5). Such a concept has a phenomenal property as its extension, but not in its character. In this sense, it is exactly the opposite of a Janusian concept. A major difference between these two concepts is that the color scientist who is trapped in a black-and-white room can have a relational phenomenal concept but not a Janusian concept related to colors.

Before closing this section, let me consider a prominent model of concepts that bears some resemblance to my proposal, i.e., the response-dependence view. According to this view, the mastery of a response-dependent concept for concept possessors is defined in terms of “their believing of anything they encounter that it is T if it seems T and there is no evidence of unfavourable influences” (Jackson & Pettit, 2002:99). Thus, assuming that red is a response- dependent concept, its mastery involves being able to infallibly track its extension in favorable conditions.

Here, I list three differences between the Janusian framework and the response-dependence view. First, a response-dependent concept is individuated based on its extension: a change in the extension results in the change in the concept. Assuming that cat is a response-dependent concept, suppose that a subject identifies cats as the extension of cat until t in the ideal state. After t, she starts identifying dogs as the extension of cat in the ideal state. This change in the subject’s identifying practice signals a change in the relevant concept. In such a scenario, given the response-dependent view, the first concept does not remain the same after t. By contrast, as explained above, the extension of a Janusian concept can change through time and in different circumstances.

Second, a response-dependent concept is not necessarily related to any phenomenal element. It merely requires “a primitive disposition” to identify the instances of the relevant concept in favourable conditions (Jackson & Pettit, 2002:99). Thus, contrary to Janusian concepts, there is no unbridgeable semantic gap between physical concepts and response-dependent concepts.

The third difference is that given the response-dependence theory, salty refers to an extension if the concept possessor has the ability to detect it in an ideal state. But in the Janusian framework, a subject has the ability to use a concept for an extension provided that he has grasped the relevant character. For instance, I can use salty to refer to what induces the taste of saltiness in my dog without being able to identify it at all. This makes for a significant difference between the two theories.²³

4 The Inducing-Condition for Janusian Concepts

In this section, I will focus on the role of inducing-condition in defining the character of Janusian concepts. I take the inducing-condition for a Janusian concept to be a condition under which the relevant phenomenological profile must be induced. Without specifying such a condition our description of the character is at best incomplete. This discussion will set the stage for the next section in which I will introduce my account of ought-judgments.

In the previous section, I defined the character of salty as “what induces the feeling of saltiness in the contextually relevant agent at the contextually relevant time”. To see the inadequacy of this definition, imagine an agent who suffers from a peculiar psychological condition in which he feels saltiness in his mouth each time that he touches a piece of paper with his hands. But he does not find paper salty when he puts a piece of it in his mouth. Is paper among the extensions of salty for him? Given our definition of the character of salty, it is! But intuitively, the answer is negative. For something to be salty, it needs to provoke the feeling of saltiness through the interaction with the taste-receptors. Therefore, the character of salty needs to be revised as follows: “what induces the feeling of saltiness in the contextually relevant agent at the contextually relevant time through the taste-receptors”.

The inducing-condition for a Janusian concept sometimes involves forming doxastic states. For instance, consider fearsome. Clearly, something can induce the sense of fear (which is a phenomenological profile) without the mediation of any doxastic state. However, sometimes, an object or a person can induce fear in us through our doxastic states. Imagine that Freddy Krueger is about to harm me and, not surprisingly, I find people who are about to harm me fearsome. Given the situation, I form the following judgments.

• (1) Freddy Kruger is about to harm me.

• (2) Freddy Kruger is fearsome.

In this case, the fact that Freddy Krueger is about to harm me induces the sense of fear in me through (1). Thus, (1) plays a role similar to the taste-receptors in the case of salty. Thus, forming doxastic states can be listed among the inducing-conditions of fearsome.

Here, we can categorize Janusian concepts based on whether their inducing-condition involves forming doxastic states or not. Some, like salty, do not have such an inducing condition. My beliefs seem to be irrelevant in the formation of the taste of saltiness in me. We can have Janusian concepts whose inducing-condition always involve forming certain doxastic states. I believe ought is an example of this kind. Finally, there are Janusian concepts, like fearsome, that have both doxastic and non-doxastic inducing-conditions. I call the first type brute Janusian concepts, the second type doxastic Janusian concepts, and the third type disjunctive Janusian concepts. I use the term “non-brute Janusian concepts” to refer to the second and third types.

An interesting feature of non-brute Janusian concepts is that their application sometimes invites doxastic criticism. Assume that I believe that a guy at the party is about to harm me. As a result, I find him to be fearsome. My friend can tell me that I ought not to fear that guy since he is harmless. He can even tell me that he is not fearsome! Thus, in the case of fearsome, the fact that something has induced fear in me does not necessarily make it the extension of fearsome. The effect must take place through a veridical doxastic state. (Note that fear can be induced without any intermediary doxastic state. In such cases, there is no room for doxastic criticism.)

The example of paranoia can be illuminating here. We can distinguish between two types of paranoid state. A person might become paranoid due to systematically forming false beliefs. The extension of fearsome for such a person might be identical with the extension of fearsome for a non-paranoid person, yet still she is constantly in the state of fear due to her false beliefs. By contrast, someone might become paranoid due to the fact that she is oversensitive to even a slim possibility of danger. In this case, her beliefs are not necessarily false. For instance, she knows very well that the chance of her dying in an asteroid impact is slim, yet this doxastic state induces fear in her. In such a case, it is correct to say that the extension of fearsome is different for such a person.

Now, we can return to our discussion of ought. The first step is to ask whether the inducing- condition of ought involves forming doxastic states. I believe there is no doubt that ought is a non-brute Janusian concept. One’s sense of moral motivation is induced at least in some cases through one’s other doxastic states. I explain my point with an example. Consider Ed, a committed utilitarian, who decides what is morally right to do by calculating what maximizes pleasure. Thus, Ed forms authoritative ought-judgments on what to do after calculating what maximizes pleasure. In such a case, the relevant properties induce moral motivation in Ed only after a mediatory doxastic state is formed.

Are there also cases in which one’s moral motivation is induced without any mediatory doxastic state? I am open to this possibility, though investigating it goes beyond the scope of this paper. For our purpose, as a working assumption, I take ought to be a doxastic Janusian concept, i.e., its inducing condition always involves forming a mediatory doxastic state.

Taking ought to be a doxastic Janusian concept helps us to accommodate the intuition that that we can and do misapply moral ought and that such misapplication invites correction. An agent can feel morally motivated to take a course of action due to the mediatory role of an erroneous doxastic state. In such a case, one can correct her by correcting the mediatory doxastic state. Such an agent is similar to a paranoid person for whom the extension of fearsome is similar to us, yet is in a state of fear due to his false empirical beliefs.

5 The Janusian Framework and Ought-Judgments

In the previous section, I discuss Janusian concepts and explained some of their features. Moreover, I argued that taking ought to be a doxastic Janusian concept has some significant theoretical advantages. It accounts for the irreducibility of ought to physical concepts; it explains how we might misapply the concept of ought; and finally, it clarifies the indexical nature of ought. In this section, my aim is to use the Janusian framework to offer a theory of ought-judgments that is in harmony with the contextualist analysis of “ought”-sentences. At the end of the third section, I listed three features that such a theory needs to incorporate:

• I. Recognizing the indexicality of ought and ought-judgments.

• II. Positing a distinction between authoritative and practically inert ought-judgments.

• III. Providing an account of the necessary connection between authoritative ought-judgments and moral motivation.

By taking ought to be a Janusian concept, we have already paved the way to accommodating the first intuition. In this section, I will try to offer a theory that satisfies the other two requirements.

Our first goal is to account for the difference between authoritative and practically inert ought-judgments. My suggestion is to explain the difference in terms of dissimilarities between the concepts of ought used in these two types of judgments. We can call these two concepts after the judgments in which they are deployed: authoritative ought and practically inert ought. Both of these concepts can be categorized as Janusian concepts for which the phenomenological profile associated with moral motivation plays an extension-fixing role. Yet they are different concepts due to their distinct characters.

Let me start with authoritative ought. My claim is that the features of authoritative ought, in particular its extension-fixing mechanism, can account for the characteristics of authoritative ought-judgments. The hallmark of authoritative ought-judgments is their action-guiding authority, i.e., that they enter the process of deliberation on what to do. This distinct authority differentiates this type of ought-judgments from practically inert ought-judgments that merely report the commitments of other agents.

The question is how to account for this distinct practical authority. In the third section, I maintained that by taking authoritative ought-judgments to be intrinsically motivational, we can explain why these judgments are practically relevant and enter the process of deliberation. I believe the Janusian framework enables us to incorporate this idea into our overall picture. Moreover, in the previous section, I suggested taking ought in general to be a doxastic Janusian concept, i.e., its inducing condition always involves forming a mediatory doxastic state. This assumption would be also true of authoritative ought. Thus, we can define the character of authoritative ought as “the properties that can induce moral motivation in me at the present time through veridical doxastic states”. Consequently, an agent’s authoritative ought picks out its extensions in virtue of their moral motivation-inducing effect on her at the present time.

Given the above account, what makes an ought-judgment authoritative is the concept of ought used in its formation. An authoritative ought-judgment reports that a property that is the extension of authoritative ought is instantiated. The idea that an agent can hold such a judgment without being motivated to act accordingly seems conceptually objectionable. If an ought-judgment does not come with a kind of motivational force, the concept of ought deployed in it does not refer to a property with a moral-motivation-inducing effect. Thus, the concept deployed in it cannot be an authoritative ought and the judgment cannot be an authoritative ought-judgment. In this way, motivational internalism is built into the content-fixing mechanism of authoritative ought-judgments.²⁴

Given the above formulation, the character of authoritative ought has in fact two indexical elements. First, it has a first-personal dimension to it that relativizes the content to the concept possessors. Second, it relativizes the content to the present time. The idea behind the first-personal dimension is that an authoritative ought-judgment is directly relevant to the deliberation process of the subject contemplating that very thought.²⁵ The reason for the second relativization of the character of ought (namely, the relativization to the present time) is that given the contextualist framework, an agent’s moral sensitivity is subject to change. By this, I mean the set of properties inducing moral motivation in me can change through time. For instance, I can be a Kantian for a while and then undergo a moral conversion and become a utilitarian. In such a case, a new set of properties induces the feeling associated with moral motivation in me. For this reason, we need to have a present-time dimension in our description of the character of authoritative ought.

Note that although the content of the authoritative ought is determined by the concept possessor and the present time, authoritative ought-judgments need not to be in first-person or constrained to the present time. Consider the following judgments:

• (1) I ought not to have done that yesterday.

• (2) Max ought not to abandon his child.

Both (1) and (2) are an authoritative ought-judgment if and only if the authoritative ought is used in forming them. The fact that the extension-fixing mechanism of the authoritative ought is sensitive to the present time, or the concept possessor does not imply that it cannot be used to make judgments about the past or a third person’s actions. When a person makes a judgment about a third person using the authoritative ought, the concept picks out a property to which the concept possessor has moral sensibility, not the other person. This is also true of other indexical terms and concepts. For instance, a term like “here” is sensitive to the place of the speaker at the present time. But we can make claims about the past using this term, e.g., “Napoleon spent his last hours here.”

Now we can focus on practically inert ought-judgments. My suggestion is that the character of practically inert ought can be formulated in the following way: “the properties that induce moral motivation in the contextually salient subject at the contextually relevant time through veridical doxastic states.” Since different properties have this effect on different agents, the extension of ought is not the same in various contexts. Even when it comes to a single agent, the extension of ought can vary through time. Thus, similar to the extension of salty, the extension of ought is relativized to both the contextually salient time and subject.

The above characterization might suggest that authoritative ought is just an instance of practically inert ought. In a context in which both the concept possessor and the present time are salient, someone might suggest, there is no difference between these two kinds of ought. This is not correct, however. I can make a practically inert ought-judgment in the context in which I am the salient subject without noticing that the salient subject is me. Similarly, I can join a conversation of two of my friends and keep using “he” to refer to the man who is the subject of the conversation without noticing that that person is actually me. The authoritative ought is not an instance of practically inert ought exactly in the sense that “I” is not an instance of any third person pronoun.

We can see that the Janusian approach can satisfy all three features listed above. First, according to this model, ought and ought-judgments are indexical. After all, all Janusian concepts are indexical by definition. Second, it can explain the difference between the authoritative and practically inert ought-judgments. Third, it accounts for the necessary connection between authoritative ought-judgments and moral motivation. Therefore, the Janusian framework has enabled us to provide a theory of ought-judgments that is in harmony with the contextualist analysis of “ought”-sentences. In addition to satisfying these three requirements, my suggested theory comes with a significant theoretical advantage. It can explain why the concept of ought cannot be reduced to any physical concepts (as explained in the fourth section).

At this stage, we can consider a case that is often used to challenge motivational internalism. The amoralist is an agent who can make moral judgments without being motivated to act accordingly. The possibility of the amoralist poses a significant challenge to motivational internalism as it shows it is conceptually possible to make a moral judgment without being motivated to act accordingly. However, we should be careful not to overstate the case for the amoralist. The data presented in the literature merely indicates that we cannot conceptually exclude the possibility of the amoralist making moral judgments. This does not imply that he can make authoritative moral judgments, i.e., moral judgments with action-guiding authority that enter the process of deliberation. It must be uncontroversial that if the amoralist did form an authoritative moral judgment, he would no longer be an amoralist. This is because, by definition, the amoralist does not take his moral judgments into account when he makes decisions. Therefore, as far as the Janusian framework is concerned, the debate about the amoralist boils down to a simple question: Can the amoralist form practically inert ought-judgments? The answer depends on our description of the amoralist. We need to distinguish between two kinds of the amoralist. The weak amoralist knows what it is like to be motivated by moral considerations, though he currently lacks any moral commitment. Thus, he has been exposed to the relevant phenomenological profile at some point in his life. By contrast, the strong amoralist is a person who has never been exposed to the phenomenological profile associated with moral motivation.²⁶

The Janusian framework can account for the ability of the weak amoralist to make ought-judgments. To grasp the character of practically inert ought, an agent needs to have been exposed to the relevant phenomenological profile, i.e., that of moral motivation. Given the description of the weak amoralist, he can very well form the practically inert ought and make practically inert ought-judgments without being motivated to act accordingly. But this is not the case with the strong amoralist. The problem is not that he would not be motivated to act according to others’ moral commitments. The main concern is that he has not been exposed to the relevant phenomenological profile, and, given the Janusian framework, he cannot have the practically inert ought.

Can the case of the strong amoralist be used against the Janusian framework? I believe not. Given the Janusian framework, the case of the strong amoralist is very similar to the color scientist in Jackson’s example. The fact that the color scientist can make judgments about colors does not show that she possesses phenomenal concepts. She is simply using concepts co-extensive with our phenomenal concepts. Similarly, the strong amoralist can use a concept co-extensive with ought to make quasi-moral judgments. Yet this should not be taken as evidence of his mastery of ought.

However, there is a case that might appear more challenging to address: Imagine an individual who is committed to a particular value system. But following the onset of depression, she loses all motivation to act in accordance with her moral judgments. Some have raised concerns regarding whether depression, even in its most severe forms, can truly alter our desires in such a manner. Nevertheless, for the sake of argument, let's assume that this scenario is possible. This is a more challenging case since unlike the amoralist, a depressed person can have her own value system.

The Janusian framework has enough resources to address this challenge. Let's assume that the individual who is currently experiencing depression judges that she ought to donate to a charity each week. We can ask her whether she feels any prospective guilt if she does not do so. If she answers in the affirmative, we can conclude that she possesses some degree of motivation, albeit not particularly robust, to take action. Conversely, if she denies, it is safe to conclude that she does not believe (in the authoritative sense) that she ought to donate to a charity each week, even if she previously held this belief before her depression onset. If one does not feel any sort of prospective guilt for not carrying out a particular action, it seems quite plausible to think that she no longer has the authoritative judgment that she ought to perform that action.²⁷

Before closing this section, it is apt to make a comparison between my suggested view and a view of moral judgments suggested by Dreier in his 1990 paper. Though Dreier’s main focus in the paper is the semantics of moral terms, he makes comments about moral beliefs based on which we can develop a theory that is rival to the view defended in this paper.²⁸ According to this rival view, the extension of a moral concept is fixed by the value system salient in the context of use. In the normal condition where the concept possessor does not “suffer from affective abnormalities”, the salient moral system is that of the concept possessor. In the abnormal cases, however, we should take the salient value system to be “the system constructed from the attitudes of the larger community” (Dreier, 1990:25).

Dreier’s view has two core features that differentiate it from my suggested theory. First, he articulates his view by using the notion of normality. Second, in his view, no phenomenological profile plays a role in the extension-fixing mechanism of moral concepts. A full comparison between the two theories requires more space than I have here. But I would like to make two points regarding this issue.

Dreier needs the conception of normality to address the problem of the amoralist as he takes the amoralist to be an abnormal case for whom the contextually relevant moral system is that of the larger community. Yet, he admits that he cannot rigorously specify or analyze the concept of normality, though he insists that we have an independent grip on it (Dreier, 1990:14). This leaves Drier’s theory, at best, incomplete. By contrast, the Janusian framework can tackle the problem of the amoralist without appealing to a concept like normality. As explained above, the ability of the weak amoralist to form moral judgments is similar to others’ ability to form practically inert moral judgments. Moreover, we have a good story about why the strong amoralist lacks the mastery of ought, which is very similar to what is missing in Jackson’s story of a color scientist who has never seen a red object. Our story does not make any normalcy-condition part of the character. It only indicates that exposure to the relevant phenomenological profile is required for the mastery of ought. Therefore, we do not need to appeal to any normalcy-condition in our analysis of the character of ought.

Another advantage of the theory developed in this paper is that it can account for the irreducibility of moral concepts like ought to physical concepts without giving up on naturalism. Dreier takes concepts like ought to be indexical, and that explains why they cannot be analyzed in terms of non-indexical physical concepts (1990:18–19). This cannot, however, fully accommodate the intuition regarding the irreducibility of moral concepts, as physical concepts can be both indexical and non-indexical. By contrast, the Janusian framework can accommodate the intuition regarding the irreducibility of ought to physical concepts.

6 Conclusion

In this paper I pursued three separate goals. First, I tried to offer a theory of ought-judgments that is in harmony with the contextualist account of moral “ought”-sentences. Second, I offered a non-expressivist account of moral judgments that is compatible with motivational internalism. Third, I accounted for the irreducibility of ought to physical concepts. To these ends, I discussed Janusian concepts and investigated some of their features. My hope is that I have shown that by using the Janusian framework to analyze ought-judgments, one can meet all three goals.²⁹