Introduction

The debate over whether robots have moral status is being fueled in part by the rediscovery of familiar theories of moral status and their application to robots. From the standard idea that moral status is tied to properties (Mosakas, 2021, Gilbert/Martin 2022, DeGrazia, 2022), to relational views that include phenomenological elements (Coeckelbergh, 2011, Coeckelbergh/Gunkel 2014), to reinventions of the Turing’s Imitation Game and other moral status tests (Sparrow, 2004; Danaher, 2020). In keeping with this spirit of rediscovery, Coeckelbergh’s latest proposal is to apply elements of John Austin’s Speech Act Theory (1962), John Searle’s Social Ontology (2010a), and Judith Butler’s Gender Theory (1988) to the moral status debate. The resulting performative view of moral status involves two main theses: (1) To make moral status claims, such as “robots have moral status,” is to perform a kind of speech act called a moral status declaration. (2) Moral status declarations create the social fact that the entity in question has moral status.

These claims have at least three implications. First, they challenge the prevailing idea that someone who says “robots have moral status” is expressing what they believe to be true. Instead, the performative view suggests that those who make moral claims, including philosophers and scientists, are “performers” (Coeckelbergh, 2023: 6), that is, people who directly alter social reality through what they say. This is related to the second and more general implication, which is a strong theoretical claim about the constitutive role that language plays in creating certain facts. Whether robots have moral status is said to depend only on “our linguistic performances”, that is, on “the ways we talk about robots and decide about their status” (Coeckelbergh, 2023: 4). Third, if only linguistic performances are required to create moral status, this implies that moral status is independent of particular properties or other epistemically objective features. That is, it could be given to anything and anyone.

Applying speech act theory to morality is not a completely new proposal.Footnote 1 However, the current debate about the moral status of robots provides a welcome opportunity to reassess the strengths and weaknesses of such an attempt. This paper thus presents a critical appraisal of the performative view of moral status. A general difficulty for the evaluation, however, is that the theoretical basis of the performative view is relatively vague. This is partly because it relies almost equally on the work of Austin, Searle, and Butler, and ignores the fact that these authors have very different, and in part incompatible, understandings of speech acts, and different ideas about the relationship between language and reality.Footnote 2 Thus, in its current form, the performative view is more of an intriguing proposal than a comprehensive theoretical framework.

Part of my critical response is therefore an attempt to fill in these theoretical gaps. To do so, I will first give a brief outline of the performative view of moral status (Sect. 1). Then I draw connections with Searle’s social ontology (Sects. 2 and 3). In other words, I am concerned with what might be called the “Searlean interpretation” of the performative view. On the basis of my interpretation, I will then raise some objections to the performative view (Sects. 4 and 5). Specifically, I will address its two main theses in reverse order. In Sect. 4, I argue that there is no plausible way to create moral status by declaration, which contradicts the second main thesis. I present this problem in form of a dilemma: On the one hand, individual instances of moral status declarations must fail because they do not conform to the conventions of moral discourse. On the other hand, it is neither necessary nor plausible to appeal to declarations to explain how large-scale collective recognition of the moral status of robots could come about. In the final sections, then, I will challenge the first thesis of the performative view. That is, I object to the idea that someone who says “robots have moral status” is performing a declaration. First, I will show that classifying moral status claims as declarations undermines the idea of meaningful moral disagreement (Sect. 5). Then, I propose that it is more plausible to interpret moral status claims as assertions (Sect. 6). That is, it is a speech act that does not involve the creation of a fact, but rather the assertion of a fact as given or true, and at the same time implies that the speaker is convinced of the truth of that fact.

The performative view

When someone claims that “robots have moral status,” we usually think that they are making a statement of fact that they believe to be true. This standard view is challenged by Coeckelbergh (2023), who suggests that making moral status claims should not be understood as reporting what is the case, but as an action that makes something the case. More precisely, the performative view of moral status involves two main propositions.

  1. 1)

    To make moral status claims, such as “robots have moral status,” is to perform a kind of speech act called a moral status declaration.

  2. 2)

    Moral status declarations create the social fact that the entity in question has moral status.

Both claims require some elaboration. I want to begin with the first, by providing a more detailed characterization of a moral status declaration.

Declarations are one of five types of speech acts, according to a taxonomy proposed by Searle (1976). Their defining characteristic is that they have the power to ‘align the world with the words’. That is, a speaker who successfully makes a declaration thereby creates a new fact that makes the propositional content of the declaration true. The standard example is the words spoken at a wedding ceremony. When a wedding officiant says to the bride and groom before her, “I now pronounce you husband and wife,” her words are not just a means of reporting that two people are now married. Rather, by saying it, she makes it the case that these two are now married. Because the sentence has been uttered, the facts of the world now conform to the content of the sentence.

According to Coeckelbergh’s performative view, claims about moral status have a similar power. Someone who says, “robots have moral status”, is not reporting a fact, or merely stating a belief that she assumes to be true. Rather, claims like this “are least also, if not only, speech acts that create the reality of the moral status” (Coeckelbergh, 2023: 3). Thus, by saying it, the speaker aligns the world with the worlds in such a way as to make the proposition of the claim true: robots have moral status. Thus, the first thesis of the performative view is that language—declarations, to be exact—plays a central and constitutive role in creating the fact that some entities have moral status.

I now turn to the second proposition. What does it mean to say that moral status is a social fact, rather than a moral fact? The performative view seeks primarily to capture “how we actually think and ‘do’ moral status by means of the performative use of language, about how we in that sense ‘create’ or ‘construct’ moral status” (Coeckelbergh, 2023: 5). It thus describes what some authors call common-sense morality, that is, pre-theoretical moral beliefs and practices (Driver 2022). In other words, the performative view is concerned with the moral rules and obligations that people accept and abide by in everyday life. Thus, Coeckelbergh offers a descriptive theory of how social norms and conventions actually emerge, rather than a normative view of what norms and conventions should exist. In addition, he offers a meta-ethical view of the meaning of moral claims.

However, Coeckelbergh also proposes a stronger reading of his proposal that is said to have normative implications. According to the deep performative view of moral status, “there is nothing more to moral status and morality than social facts created by speech acts and other performances” (Coeckelbergh, 2023: 5). Morality, in this “deep” view, can be reduced entirely to a set of social conventions, norms, and rules. As such, it is relative to particular communities or cultures. The deep view thus implies that social practices that have been declared morally right within a given community at a specific point in time are the morally right practices in this community. They can be changed by the people within the community, but they cannot be judged from a superior, external moral standpoint.

In sum, both the weak and the strong version of the performative view attempt to describe the empirical reality of how people attribute moral status to others, and how moral status claims should be understood. The weak version, however, leaves aside the question of whether the associated beliefs and practices conform to what external normative theories prescribe. The strong (or deep) version, on the other hand, endorses a form of moral relativism according to which what is morally right depends only on what is accepted in a particular community at a particular time.

The performative view offers an intriguing proposal for a meta-ethical debate about the meaning of moral claims. Further, it proposes a framework that explains how new moral norms emerge, with possible implications for the moral status of machines and other entities. However, it leaves a number of important questions unanswered. For example, why do moral claims have the power to create new facts? What is the mechanism that allows a “mere” utterance of words to have such a profound effect on reality? In short, how can the use of language create a fact out of thin air? The references to Austin, Searle, and Butler are insufficient to provide a comprehensive theoretical frame, particularly as these three authors hold very different views on the nature and function of speech acts.

In order to provide a critical assessment of the performative view, I will therefore begin by offering an interpretation of it that attempts to asnwer the questions I have just outlined. For this purpose, I turn to the work of Searle, who has written extensively on the connection between language, speech acts, and social reality, to offer a Searlean interpretation of the performative view. I take this to be a plausible and charitable interpretation, since Coeckelbergh makes explicit and extensive reference to Searle’s work on social ontology. Therefore, I assume that my attempt to fill in the theoretical gaps does not change Coeckelbergh’s proposal, but rather elaborates on it. I recognize that this is only one possible interpretation, and that other people may try to advance an Austinian or Butlerian interpretation of the view. But that is beyond the scope of this paper. In the next section, I will give a brief outline of Searle’s social ontology, before presenting the connections to the performative view in Sect. 3.

A brief outline of Searle’s social ontology

Searle’s (2006, 2010a, b) social ontology attempts to reconcile two seemingly contradictory claims. One the one hand, social facts exist only by virtue of collective acceptance or recognition or acknowledgment. They are ontologically subjective, meaning that they exist only insofar as people believe that they exist. On the other hand, social facts are facts, and not matters of opinion. Thus, although they are ontologically subjective, they are epistemically objective. For example, it is not a matter of individual beliefs or opinions whether a given piece of paper is a $20 dollar bill or not, “it is an actual objective institutional fact” (Searle, 2018: 53). Searle’s work seeks to explain how this social reality can emerge in a world that, at the lowest level, consists only of physical particles (Searle, 2010a, b). For the purposes of this paper, I will only attempt to give a brief overview of this complex theory that Searle has been working on for much of his career.

First, note that Searle (2006) distinguishes between three kinds of facts. Brute facts, that is, the facts that the natural sciences describe, are ontologically objective. That is, their existence does not depend on any form of recognition. By contrast, social facts emerge from the “collective intentionality of two or more human or animal agents” (Searle, 2006: 17). Thus, social facts are observer-dependent; they wouldn’t exist if they weren’t the content of collective beliefs, desires, goals, and so on. Institutional facts are a subclass of social facts. Both emerge from being collectively represented as existing, but only the representation of institutional facts requires language.

To explain the relationship between language and institutional facts, one must first understand the concept of a status function. According to Searle, every institutional fact involves the collective imposition of a status function on some entity. This allows that entity to perform a specific role or task. Importantly, a status function is not derived from the physical properties of the entity it is assigned to. For example, using a stone as a hammer imposes a function on the stone, but this is not a status function, because the function is tied to the physical properties of the stone. By contrast, an entity that can perform a status function can do so only because people have collectively agreed that it can.

An obvious example is money. The piece of paper in my hand, unlike the knife in my pocket, does indeed perform a function, but it performs the function not in virtue of its physical structure but in virtue of collective attitudes. The knife has a physical structure that enables it to cut and perform other knife-like functions, but money has no such physical structure. The physical structure is more or less irrelevant, provided only that it meets certain general conditions (such as being easy to recognize as money, easy to transport, hard to counterfeit and so on). (Searle, 2006: 17)

The general form of status functions is “X counts as Y in context C”. To become an institutional fact, it must be collectively accepted that X counts as Y in context C. For instance, a piece of paper issued by the Bureau of Engraving and Printing (X) counts as a $20 dollar bill (Y) in the United States (C) (Epstein, 2023). Or, a person elected by the German Bundestag (X) counts as a Bundeskanzler (Y) of Germany for the duration of four years (C). According to Searle, all human institutions are matters of status functions. This can be summarized in the following equation:

1) Institutional facts = status functions.

Searle (2006, 2010a, b) believes that the collective imposition of a status function requires language. Since state functions are not tied to physical properties, they require that something (X) be represented as something else (Y). For example, a piece of paper must be represented as having a value of $20. According to Searle, this is only possible by means of linguistically structured representations. Thus, “[the] form of the collective acceptance has to be in the broadest sense linguistic or symbolic because there is nothing else there to mark the level of status function” (Searle, 2006: 20). In other words, since status functions are not rooted in anything physical, they must be constructed linguistically.

Furthermore, Searle assumes that language is necessary in the creation of new institutional facts. More precisely, institutional facts are created by successfully executed declarations. To repeat, a declaration is an utterance that makes its own propositional content true. For example, by saying “You are fired,” my boss can make it a fact that I am fired. Searle (2010a, b) believes that all institutional facts are assignments of status functions, and that all status functions have the underlying logical structure of declarations.Footnote 3 That is, a new fact comes into existence by representing it as already existing. In some cases, like marriage, this involves the actual performance of a speech act. In other cases, it can only be a “maneuver” which has the logical form of a declaration (Prien et al., 2010). Thus, we can make another addition to the equation:

2) Declarations → institutional facts = status functions.

In short, declarations create new institutional facts, i.e. they impose a status function on some entity or type of entity.

The final element of Searle’s theory that I want to highlight is the idea that imposing status functions creates deontic powers. Deontic powers include rights, duties, obligations, requirements, responsibilities, authorizations and permissions. These kinds of powers arise from and are tied to the collectively recognized institutional functions. For example, assigning someone the role of university professor creates new rights and obligation. This is true both for the person to whom the role is assigned and for others who may have new obligations or claims against him or her. Similarly, declaring a piece of paper to have the status of money creates a system of rules around it that does not exist around ordinary pieces of paper. In other words, status functions are vehicles of power in that they both create and structure the power relations that govern the social-institutional world (Searle 2006). So, I will make one final addition to the set of equations and implications:

3) Declarations → institutional facts = status functions → deontic powers.

This is, of course, an incomplete summary of Searle’s social ontology, but I will leave it at that. It is sufficient for the present purpose, which is to make some additions to the performative view. Before doing so, however, I would like to briefly summarize the most important points. (1) Searle social ontology focuses primarily on institutional reality. (2) Institutional reality is ontologically subjective, but epistemically objective. (3) Each institutional fact involves the collective assignment of a status function, which confers deontological powers. (4) The collective assignment of status functions requires linguistic representation. (5) New institutional facts are created by declarations, or maneuvers of the same logical form. In the next section, I will present the Searlean interpretation of the performative view.

The Searlean interpretation of the performative view

Based on the elements of Searle’s social ontology presented, I propose a slight modification of the performative view. As I see it, the performative view of moral status is concerned with institutional reality rather than social reality of moral status. Some people might object to this suggestion on the grounds that morality is not institutional in nature. However, authors like Beauchamp and Childress (2001: 3) describe common-sense morality, which is the phenomenon the performative view seeks to explain, as “a social institution, [which] encompasses many standards of conduct, including moral principles, rules, rights, and virtues.” And Searle (2010b: 39) himself highlights a number of parallels between ethics and institutional status functions, most notably that ethics requires language and creates duties and obligations. These parallels can also be found in the performative view, as I will show below.

First, Coeckelbergh’s concept of moral status corresponds well to Searle’s concept of a status function. Recall that status functions are imposed by means of collective acceptance and are not derived from any physical properties. In a similar fashion, Coeckelbergh (2023: 3) rejects the assumption that moral rights are based on physical properties; he believes that moral reality can never “be entirely separated from the way we know and socially construct it through language and otherwise.” Furthermore, the assignment of a status function structures relationships of rights and obligations. Similarly, granting moral status to, say, a robot implies that other people have obligations to that robot, and that the robot may have legitimate claims against others.

Second, Coeckelbergh (2023: 2) claims that “language is a condition of possibility for moral status ascription,” echoing Searle’s assumption that language is a prerequisite for institutional reality. In doing so, the performative view makes explicit reference to the idea that moral status is created by means of declarations: “Via the performative use of language, in the form of declarations, we create a particular moral status in a specific context.” (Coeckelbergh, 2023: 3). While Searle’s theory holds similar views on institutional facts, it differs when it comes to social facts. Social facts, in Searle’s (2006) view, arise from collective intentionality alone, which need not be linguistically represented. Thus, a pack of wolves hunting together create a social reality by means of their non-linguistic coordinated action. Institutional reality, on the other hand, is necessarily represented linguistically. As already mentioned, it involves the imposition of functions that are not derived from physical characteristics, but depend only on collective attitudes, and as such, can only exist in linguistic form (although see Hindriks, 2013).

If the moral status of, say, a robot does not depend on physical properties, but exists only through collective linguistic representation, then, following Searle, the robot’s moral status is an institutional fact. Thus, the second main thesis of the performative view needs to be modified as follows. I believe that this does not change the underlying claims, but it does tie them together more coherently.

2*) Moral status declarations create the institutional fact that the entity in question has moral status.

Thus, the granting of moral status involves the collective assignment of a status function by declaration or an equivalent maneuver. The general form of a status function is “X counts as Y in context C”. Leaving the context aside for now, what do X and Y stand for in this case? A straightforward way to fill in the Y-variable would be to say that “X counts as morally considerable” or “as morally valuable”. A more interesting question is the specification of X. For it can either be filled by an individual, such as the Boston Dynamics robot Spot, or by a type, such as all zoomorphic robots. Following Mary-Ann Warren, I will assume that X represents a type of entity. Warren (2000: 9) notes that “ascriptions of moral status serve to represent very general claims about the ways in which moral agents ought to conduct themselves towards entities of particular sorts.” Thus, they are usually ascribed not no specific individuals, but to members of a group. In this sense, to declare that robots have moral status seems impose as a status function on all robots, or a particular type of robots.Footnote 4

The dilemma of the performative view

Highlighting the parallels in Searle’s and Coeckelbergh’s views has provided some insight as to why attributions of moral status require language. But still, a major question is left unanswered: How is it that moral status declarations have the power to change reality? As was mentioned, Coeckelbergh’s performative view does not provide an answer. It points to the general idea that language creates reality and alludes to these ideas in the work of Austin, Butler, and Searle. But it does not commit itself to any kind of explanation of how the utterance of certain words can have an effect on reality. I think that this failure to address the link between language and reality is telling, because it is here, that the critical problem of the performative view lies.

This problem can be expressed in the form of a dilemma. According to Ludwig (2017), there are two different ways in which declarative speech acts can be used. In the first case, there are pre-existing conventions that ensure the success of the declaration. In the second case, declarations are offered provisionally and succeed if they are collectively accepted. The dilemma arises because none of these cases provides a plausible explanation of how moral status is assigned. I will argue that, in the first case, moral status declarations will necessarily fail because they do not correspond to the conventions that govern moral discourse. In the second case, we do not need to appeal to declarations to explain how people arrive at collective beliefs about moral status. My objections are inspired, in part, by the criticisms that Ludwig (2017), Hindriks (2013), Fotion (2011) and Prien et al. (2010) have raised against Searle’s social ontology, as well as his institutional view of Human Rights (Searle, 2010a).

Moral status declarations in pre-existing conventions

I begin with the first horn of the dilemma. In this case, there are conventions in place “relative to which the utterance of a sentence that asserts that a thing has a certain status function (perhaps by someone with a particular status role) is collectively accepted as sufficient for the thing henceforth to have that status function” (Ludwig. 2017: 182) For example, it is collectively accepted that the U.S. congress declares the winner of the 2020 presidential election. The reason why this declaration succeeds depends upon “antecedent conventions, which are not merely linguistic conventions, and which make a place for them” (Ludwig, 2017, 182). According to these antecedent conventions, Donald Trump, for example, did not have the collectively accepted authority to declare the winner of the presidential election. As a result, the same words spoken by him did not have the intended effect.

This highlights the fact that the power of declarations does not lie in the words, but in the conventions that govern the use of the words. As Searle (1989: 555) describes it, there is no special semantic property of performativity, “[w]hat we find instead are human conventions, rules, and institutions that enable certain utterances to function to create the state of affairs represented in the propositional content of the utterance.” More precisely, there are four necessary conditions that must be met for a declarative speech act to succeed in changing institutional reality according to Searle (1989: 548):

  1. (1)

    There is an extra-linguistic institution.

  2. (2)

    The speaker, and sometimes the hearer, occupy a special position within the institution.

  3. (3)

    There is a special convention that certain literal sentences of natural languages count as the performances of certain declarations within the institution.

  4. (4)

    The intention by the speaker in the utterance of those sentences that his utterance has a declarational status, that it creates a fact corresponding to the propositional content.

For a declaration of moral status to be successful, all four conditions must be met. I will leave aside the question of whether there is such a thing as an institution of morality or moral status. Regardless of what one believes about this question, I think the main difficulty for the performative view arises from conditions two and three. These require that there be a collectively accepted convention that ensures that a speaker with appropriate authority can create new facts by uttering certain words. Again, the paradigmatic example is the case of a marriage officiant whose words produce a marriage.

Correspondingly, a moral status declaration is successful if (1) it is uttered by a speaker of a special position, and (2) there is a convention according to which saying “robots have moral status” makes it a fact that robots have moral status. But it seems to me that both are implausible. First, there is no special position with respect to moral status assignment that would allow any particular speaker to declare that robots have moral status. Unlike, say, some religious traditions or the law, common-sense morality has no real authority figures. Evidently, there is no single speaker who can single-handedly make it a fact that an entity has moral status. If my neighbor tells me that toasters have moral status, I have no reason to abide by his words, since my neighbor does not have the authority to impose moral obligations on others. On the other hand, if everyone were equally entitled to make that declaration, then no such fact could ever arise, since there is moral disagreement on almost any subject.Footnote 5

To illustrate this point, allow me to make a brief detour into animal ethics. I ask the reader to bear with me at this point, as I believe this additional example will help make the matter clearer. According to Coeckelbergh, the performative view enables us to rethink why Peter Singer’s book Animal Liberation (1977 [1975]) has been so socially influential:

When Singer wrote his Animal Liberation, for example, he framed this argument in a realist way: animals have the capacity for suffering, and therefore they have a moral status. But his utterances […] also helped to change the situation of some animals […]. The words mattered politically; they were politically performative. (Coeckelbergh, 2023: 8)

Thus, the performative view suggests that even statements that seemingly express a statement of fact, or a belief about such a fact, can play the role of declarations. But based on the four necessary conditions, I do not think this view can be sustained.

First, although Animal Liberation made Singer a prominent public voice, it did not give him the collectively accepted authority to single-handedly declare moral facts. After all, there were other influential philosophers at the time, such as Raymond G. Frey (1979), who argued that animals had no moral status, and why should they not have equal authority to declare the opposite? Unlike a judge who can give a legally binding verdict, or a head of state who makes decisions that people must obey, philosophers and other participants in moral debates have no comparable capacity granted to them.

Second, according to the conventions of moral discourse, it is decidedly not enough to simply state that “animals have moral status.” As DeGrazia (2008) points out, claiming that an entity has moral status without being able to explain or justify why this is the case is nothing more than dogma, and therefore without effect. The reason Singer’s work is influential is that he presents a compelling argument for why animals are morally important. If this were not the case, then the claim that “animals have a moral status” would not have had the same effect.

Furthermore, the fourth condition states that the speaker must have the intention to make a declaration in order for the declaration to be successful. This rules out cases in which the speaker “accidentally” makes a declaration when his intention was actually to convince the audience of the truth of some claim. I will not speculate here about what Singer’s motives were when he wrote Animal Liberation. But if his intention was not to declare the status of animals directly, then he also did not do so by accident. Similarly, if robot ethicists like Joana Bryson “deny that they co-create the moral status realities they talk about,” as Coeckelbergh (2023: 8) notes, then on the basis of the fourth condition we cannot say that they create that reality anyway. To successfully make a declaration, the speaker must intend to make the declaration.

Finally, I think we cannot even be sure that the moral status of animals has ever achieved the status of institutional fact, in the sense of being collectively accepted as existing. As Coghlan (2023) notes, Animal Liberation has “changed the world but not enough”. Although some countries have introduced important changes in legislation, meat consumption is on the rise and still largely exposes animals to cruel conditions. Against this background, it may seem inadequate to speak of the moral status of animals as an institutional or social fact, especially compared to the certainty with which we agree that a particular piece of paper is a $20 dollar note, that Olaf Scholz is the German Bundeskanzler, or that two people are married.

Undoubtedly, Animal Liberation has had an impact on moral beliefs about animals, at least in Western society. But the fact that words have had an impact does not mean that they have the status of declarations. There is a significant difference between saying that someone’s words changed public opinion through persuasion and saying that someone’s words directly created an institutional fact. Language is part of both processes, but the role of language plays in each process is very different. In the case of declaration, a single utterance of the right words by the right person in the right context is sufficient to create a new institutional and thus collectively accepted fact. If the four conditions are met, saying the words is equivalent to creating a fact. In contrast, to reach a collective agreement through persuasion, it is not enough that certain claims are uttered, such as that animals have moral status. Rather, what is important is the content of the claims and the way in which those claims are justified. Similarly, a collective acceptance of the moral status of robots will not result from single or repeated declarations if these cannot be justified. Whether or not robots will have moral status depends primarily on whether humans can make a convincing case that they should.

To summarize, I think the first horn of the dilemma is this: Given the norms and conventions that govern moral discourse more specifically, any attempt to declare moral status would fail. First, because no single person has the authority to impose new moral obligations on others. Second, if everyone had the same authority, then it is unlikely that a consensus would emerge. Finally, according to the conventions of moral discourse, moral claims are accepted only to the extent that they can be justified. Thus, simply saying that robots have moral status will have no effect.

Moral status declarations without pre-existing conventions

The second way in which declarations can be used is to establish a status function ‘out of thin air.’ Thus, the declaration is made in a context where no prior arrangements guarantee its success (Ludwig, 2017). A common example of such cases is the invention of a new game, in which a group of people must agree on certain rules in order to play, without having pre-established conventions that guide the process. Thus, we are concerned with what may be called “original declarations.” Applied to the performative view, the idea seems to be that all members of a society must somehow, collectively, come to an agreement about how to treat entities, such as robots, without being able relying on already established roles and convention.

Ludwig (2017: 182) notes that, in cases like this, a declaration is always offered provisionally, and only succeeds if it is collectively accepted. Without an established institution with its roles and rules, there is nothing that ensures the success of the declaration. Therefore, the utterance of an original declaration has the status of a proposal, leaving open whether it will be accepted or not.

I believe that this interpretation may correspond best to what Coeckelbergh has in mind, when he says that the moral status of robots and AI is co-created by all of us.

[…] what we say about robots and AI, as users but also as stakeholders in processes of development and policy, is not only about what robots and AI are but also about what they could and should become. By means of speech acts and other acts, we performatively create their reality and their status. (Coeckelbergh, 2023: 8)

The envisioned scenario seems to be something like this: In a given society, it is not yet a collectively recognized fact that robots have moral status. To create this institutional fact, there must first be a provisional offer of a general status function of the form “robots count as morally significant.” This tentative offer can be made by any member of this society, as there are no pre-determined roles of authority. Thus, anyone who says the words “robots have a moral status” is trying or offering to declare that robots have a moral status. Whether their declaration is successful depends on whether people collectively accept this proposal. Once the status function is collectively recognized, it is an institutional fact that robots have moral status.

But even this account runs into a problem: In oder to explain how large-scale collective agreements come about, it is not necessary to appeal to declarations, as several authors have already pointed out in response to Searle’s social ontology (Ludwig, 2017; Hindriks, 2013; Prien et al., 2010; Blackburn, 2010). What is more, the idea is even implausible in many cases. For example, it does not seem very likely that institutions such as money, governments, the family, and so on arose because, at some point, someone performed speech act. Instead, as Blackburn (2010) puts it, conventions, norms, and the imposition of status functions “can be slow growths arising in our lives together through recognition of the mutual advantages to be gained […],” which is why “[d]eclaration is unnecessary […].” In his earlier writings, even Coeckelbergh (2014: 75) himself defended the view that who belongs to the moral community “cannot simply be declared or agreed upon” but “has to grow.” That is, collectively accepted institutional facts can emerge from nonverbal interactions and cooperation, they can be enforced by emotional responses, they can be guided by a deep-seated sense of fairness and reciprocity, and so on.

Searle (2010a) has responded to objections of this kind by saying that there does not necessarily have to be an actual declaration, but something must happen that has the logical form of a declaration. For example, for robots to have moral status, it must be collectively and linguistically represented that they have moral status. And this representation, Searle argues, has the logical form of a declaration. However, this seems to be a much weaker claim than the initial idea that all institutional facts are created by means of a declaration. First, it is not entirely clear what an act or maneuver of the same logical form as a declaration would be (Ludwig, 2017, (see also Hindriks, 2013, Prien at al. 2010). The declarations that are discussed in the literature are for the most part cases of actual utterances, either written or spoken. The claim that other maneuvers or linguistic representations can have the logical form of a declaration is not only extremely vague, but it also seems to undermine the idea that language plays a strong constitutive role for institutional reality. For, if all that is meant is that our beliefs about institutions involve linguistic representations, then “the claim that language is constitutive of institutions is trivially true” (Hindriks, 2013: 381). Yes, we use language to think and talk about social issues, such as whether robots have moral status. But if this is what the performative view suggests, then the claim that moral status is ‘linguistically-performatively constructed’ becomes so all-encompassing that it is not very interesting.

More importantly, there may still be non-linguistic factors that cause us to think and speak differently about robots. For example, the fact that we have discovered that they have a capacity that we did not know about before. What the performative view seems to imply is that nothing more than language, nothing more than a declaration and its collective acceptance, is needed to create a new institutional fact. But there seem to be a variety of other ways in which the collective assignment of status functions can come about.

To sum up, the performative view is confronted with the following dilemma: On the one hand, individual instances of moral status declarations must fail because they do not conform to the conventions of moral discourse. On the other hand, it is not necessary, or even implausible, to appeal to declarations to explain how a large-scale collective recognition of robot moral status may come about.

Considerations on moral disagreement

In the beginning, I have stated that the performative view of moral status involves two main propositions.

  1. 1)

    To make moral status claims, such as “robots have moral status,” is to perform a kind of speech act called a moral status declaration.

  2. 2)

    Moral status declarations create the institutional fact that the entity in question has moral status.

The discussion so far has focused on the second proposition. In these final sections, I will take a critical look at the first proposition. That is, I want to object to the idea that claims about moral status can be classified as declarations in the first place. The problem of this first proposition is that is undermines the notion of meaningful moral disagreement.

By meaningful moral disagreement, I mean that two people with different moral opinions can engage in a debate in terms of reasons and justifications. Imagine Susan and Donald, who come from different communities, meet. In Don’s community, it is common practice to treat robots kindly, and any unfriendly behavior towards them in frowned upon. In the Susan’s community, robots are considered to be mere machines, and no special consideration is given to them. According to the performative view, it is an institutional fact of Don’s community that robots have moral status, and an institutional fact in Susan’s community that robots do not have moral status. Coeckelbergh explicitly endorses this form of cultural relativismFootnote 6, and rightfully so, since he attempts to capture the empirical realities of diverse moral practices and beliefs.

What his view fails to capture, however, is that despite the fact that different cultures have different moral beliefs, there can still be cross-cultural moral debates that involve justifications and reasons for these beliefs. Debaters from different cultural backgrounds can, at least to some extent, explain what underlies the moral practices of their community—whether it is shared beliefs, emotional reactions, religious convictions, and so forth. They can make transparent the values they hold dear and the hierarchy in which they place them. Some moral disputes can even be resolved if both parties discover that they accept a shared principle. But the performative view seems to leave no room for the kind of moral debate that most of us, I would argue, are familiar with. For what could Don say to Susan to convince her that, she too, should treat robots kindly? All he could say was that in his community, robots have been declared to have moral status, and therefore it is a fact that they do. No justification is required for the declaration to be successful. Therefore, neither Don nor Susan could explain to each other why robots have (not) been declared morally considerable in their respective communities.

Strictly speaking, the performative view might actually not leave room for the notion of a debate at all. For two people wo make opposing declarations do not seem to be engaging with each other, rather, they seem to be testing the limits of their authority. Think of Donald Trump’s attempt to overrule the decision of the Congress with respect to the result of the 2020 election. Or imagine a situation in which a mother declares that her newborn child is called Mary, while the father proclaims that her name is Jane. According to the performative view, two people making differing moral claims find themselves in similar situation with no possibility of resolving the debate. This result is particularly troubling in the case of cross-cultural decisions that require shared principles, as is so often the case in the context of emerging technologies. In this sense, the performative view seems to offer a rather pessimistic outlook for future cross-cultural ethical guidelines.

In sum, the proposal that moral status claims are declarations does not seem to capture the conventions that govern moral discourse and the importance of giving reasons and justifications in this context. In the absence of such meaningful debates, the resolution of cross-cultural ethical conflicts seems untenable. In the remainder of this paper, therefore, I will reconsider whether claims about moral status are really declarations. I suggest that, on the basis of Searle’s speech act theory, it may be more plausible to interpret them as assertions.

Why declarations?

Applying speech act theory to moral status claims does not require us to conclude that they are declarations. On the contrary, it is a central assumption of speech act theory that two utterances of the same sentence in different contexts can constitute different actions. For example, “I’ll be there” can be both a promise and a warning, depending on the speaker’s tone, facial expression, body language, and so on. Speech act theorists refer to this as the “illocutionary force,” of a sentence, or the “illocutionary act” that is performed in uttering the sentence, which is independent from the propositional content of the sentence. In many cases, the context makes it clear which kind of action is performed (Searle, 1965).

Thus, the utterance “robots have moral status” can encompass various illocutionary acts. In the context of moral debates, the most plausible interpretation, in my view, is that they are assertions. Assertions are characterized by the speaker’s commitment to the truth of the proposition they utter (Searle, 1976). Thus, the utterance is consistent with the speaker’s underlying intention to communicate his genuine belief and to persuade the listener. This perspective, which I argue is more advantageous than the performative view, addresses several of the issues outlined above. In other words, the assertion model of moral claims overcomes the problems faced by the declaration model.

First, anyone can make assertive statements, just as anyone can express moral claims. This is because the requirements for a successful assertion are less demanding than for a successful declaration (see Searle 1976 and 1989). To successfully make an assertion the speaker need only adhere to the linguistic conventions. Unlike a declaration, an assertion does not require that the speaker occupies a special role of authority. Thus, the assertion model accommodates the inclusive nature of anyone making assertions, which corresponds to the accessibility of moral claims. Second, classifying moral status claims as assertions better captures the intentions of people wo participate in moral debates. According to Searle (1976), a speaker who makes an assertion wants to communicate what he believes to be true and to convince the hearer of it. From what I have observed, this is exactly what the people who are involved in the moral debates are trying to do. Thus, the assertion model better accommodates the objectives in moral discussions. Third, highlighting the goals of moral debates underscores the importance of reasons and justifications, which are integral to persuasive discourse. In order to convince others that what we are saying is true, we must provide support. And while persuasion involves more than making a convincing case, it remains an essential aspect. Finally, this viewpoint aligns with commonly agreed-upon practices in moral discussions, recognizing the widespread belief that expressions about moral status describe factual truths about the world. As Coeckelbergh (2023) notes, it is “the usual view that moral status utterances are a description of truths in the world (in particular, truths about entities based on truths about their properties).” Importantly, if we consider moral status claims as assertions, it opens the door for genuine moral disagreements. This becomes possible because people can express their beliefs persuasively by offering reasons and justifications, all of which essentially falls within the realm of making assertions.

Conclusion

This paper explored Coeckelbergh’s performative view of moral status, drawing parallels with Searle’s social ontology. This analysis has highlighted two central propositions of the performative view: (1) To make a moral status claim is to perform moral status declaration. (2) A successful moral status declaration creates the institutional fact that the entity in question has moral status. Closer examination revealed that both claims are flawed. The second proposition was shown to face a dilemma: On the one hand, individual instances of declaring moral status are bound to fail because they do not conform to established conventions of moral discourse. On the other hand, it is both unnecessary and implausible to rely on declarations as an explanatory framework for elucidating the mechanisms behind widespread collective recognition of moral status. With respect to the first claim, it was shown that accepting it undermines the possibility of meaningful moral disagreement. As a remedy, this paper proposed a shift in perspective: interpreting moral status claims as assertions rather than declarations. As such, they are a type of speech act that asserts a fact while implying the speaker’s conviction. This refined approach offers a more plausible lens for understanding moral status that avoids the pitfalls of the performative view.