Navigazione – Mappa del sito
Discrimination in philosophy

Trust, Agency and Discrimination

Jacopo Domenicucci
p. 83-102


This paper attempts to clarify the relations between trust, agency and latent forms of discrimination. Its main argument is in social philosophy, and it articulates considerations from moral psychology and the philosophy of language. The aim is to provide a better insight into the deflated level of trustworthiness that members of stigmatized categories are credited with. Identity prejudice determining low levels of credibility for its victims has received substantial philosophical attention in the wake of the epistemic injustice literature. Instead of the standard cognitive concept of trust usually brought in, I apply a moral (or agential) account of trust (drawing on Domenicucci and Holton 2017) to discriminatory issues. We can fail to trust the trustworthy (Jones 2013) in a variety of ways. When it comes to the members of discriminated categories, I argue in the first part, this happens because they are refused a necessary condition of trust: a presumption of agency. This presumption of agency, prima facie attributed to human adults, is precisely what goes wrong for the members of stigmatized groups. Not recognised as individual agents, but as mere quasi-agents, the discriminatees cannot be trusted: their trustworthiness actually becomes unthinkable. The very possibility to give reasons for or against out-group members’ trustworthiness is thus dismissed. Articulating documentality and feminist philosophy of language, the second part shows that members of stigmatized categories have their conduct described and recorded through a non-agential grammar. Once widespread and socially dominant, non-agential descriptions of their action give the discriminatees a peculiar moral status that jeopardizes the very possibility of their being recognised as trustworthy.

Torna su

Termini di indicizzazione

Torna su

Testo integrale

  • 1 For a review of the sociological evidence, see Smith 2010: 453-475.
  • 2 Jones 2013, focusing on situations where we fail to trust the trustworthy, and Hawley 2014, focusin (...)

1The relations between trust, agency and discrimination are in need of clarification. Suffering a systematic lack of trust is commonly seen as part of being the member of a socially stigmatized category1, especially in a modern liberal society that recognizes legal status equality and where discrimination is not deliberately legally or politically implemented. Hence, it seems reasonable to expect of a theory of trust, that it can be applied to clarify the nature of the negations of trust at work in discrimination. However, little has been done in this direction in analytic philosophy2. Standard social or economic accounts of trust rarely prove fruitful in this direction and can hardly distinguish discrimination from mere partiality in the allocation of resources or social goods. The only concept of trust extensively applied to discriminatory issues is that of epistemic trust, explored by the epistemic injustice literature (especially Fricker 2007, Haslanger 2014, Pohlhaus 2014 and Congdon 2017). Though allowing for a rich insight into testimonial injustice and related phenomena, this epistemic approach only highlights a single specific aspect of trust deficits suffered by discriminatees. Here, taking a closer look at the imbrication of trust and agency, I suggest that a non-merely epistemic concept of trust helps describe what goes wrong when we fail to trust the members of socially stigmatized categories. The aim is to locate what the debate on epistemic injustice has found in the broader picture. This should help highlight the moral harm of discriminatory trust deficits.

1. Agency, quasi-agency and trust

i) Two approaches to quasi-agency, and the way they can diverge

  • 3 By agent, we mean «one who or that which acts; as opposed to the patient, or the instrument […], th (...)

2Trust, as argued in Holton and Domenicucci (2017) is a cluster concept that is best understood in a dyadic way: there is a primacy of dyadic trust, i.e. trust whose object is a person, and then this relation can support a variety of three-place acts. Three-place trust acts do not stand on their own, they are grounded in a practical stance that we can take toward somebody we take to be a person, i.e. an autonomous and responsible agent. Being an agent, being treated as an agent, is thus a necessary (clearly non-sufficient) condition for being trusted. For any entity to be a possible object of trust, it has to be treated as an agent3.

3This apparently simple necessary condition, I argue, plays a central role in the specific trust deficits suffered by members of stigmatized groups, i.e. in the fact that members of discriminated groups only enjoy deflated levels of trust. Focusing on the way trust is bound up with agency, I propose a simple account of discriminatory distrust, which does not need to climb up to the moral heights of “personality” and “respect”, but sticks to the more basic notion of agency. Clarifying what it is to be taken as an agent, should help us understand what can go wrong for the victims of discrimination, and how they can be turned into something which cannot be treated as as worthy of trust as dominant group members.

4We sometimes treat inert objects as agents, for example in superstition, and we sometimes treat agents as passive entities. Note that this divergence is not binary. It is not just a matter of failing to treat passive entities as passive entities and agents as agents. Divergence between intrinsic and attributed agency for the same entity is not all-or-nothing in principle. It is rather often a matter of degree: agency can be over-attributed or under-attributed.

5Let me call pseudo-agents the entities which lack most of the intrinsic criteria for agency but that are viewed, deemed and treated as agents, and quasi-agents the entities which do display the intrinsic sufficient conditions for agency but whose agency is not recognised to the appropriate degree. Pseudo-agents are pseudo in the sense that they somehowlook like agents: they are not metaphysical agents but are mistakenly taken as -or even “pretend to be” (Ferraris 2007: 21, our translation) – agents. Quasi-agents are quasi in the sense that they are proper metaphysical agents but they fail to be appropriately recognised.

  • 4 A limit case where she actually produces effects but no one notices, as they affect nothing and nob (...)
  • 5 Note that the notion of quasi-agent is not coextensive with Ann Cahill’s «derivatized subject». As (...)

6A quasi-agent is not really seen as the doer of her action. She actually produces effects in the world, and inquiry into her metaphysical nature would cast her as an agent, but she is not the subject to whom the action is attributed. Other causal chains enter the story and external factors interfere with a fully-fledged attribution of agency. She is taken to be a secondary cause, or even an occasional cause, to use classic metaphysics’ terms. She is generally seen as taking part in the events, as contributing to the outcomes4, but not as being able to initiate action. Before being a co-cause, she is herself an effect. She cannot really act or speak on her own behalf, as her action will often be perceived as a consequence of something else5.

ii) Presumption of agency for humans

  • 6 Gell 1998 develops an agential account of art and Ferraris 2007: 193-202 talks about artworks as « (...)
  • 7 See Dennett (1971: 104), on a parallel line, about incorrigibility for persons: «For intentional sy (...)

7Pseudo-agency received conceptual scrutiny in anthropological and philosophical theories of art, especially Gell 1998 and Ferraris 20076. In this paper, we focus on the notion of quasi-agency, which is relevant, as it should appear, for discriminatory issues. Quasi-agency raises distinctive issues for each form of agency. Let us turn to human agents. Humans display the following property: they know they are agents themselves, they know their average adult peers also tend to be agents, and they communicate about their actions and the attitudes surrounding them, making their agency explicit and manifest7. These practices ground reasonable expectations of agency when meeting a fellow human adult.

  • 8 I take it that, when one has a gradual concept of agency as I do, responsibility and autonomy imply (...)
  • 9 Cfr: Dennett (1971: 93) goes further on this line: «One fact so obvious that it is easily overlooke (...)

8The idea of a presumption of agency for humans fits well the Strawsonian picture of «ordinary reactive attitudes» (Strawson 2013: 343). “Ordinary” is relevant in stressing that our default posture toward fellow humans is the participant stance, the one presupposing that one is facing an autonomous and responsible agent8. Thus, each of us expects (whether this is or not justified is another question) to be treated as an agent. There is a reasonable presumption of agency for adult human beings9. Human adults are generally presumed to be agents, that is, they are individually attributed a high level of agency prima facie (whether they are seen as well or ill-intentioned is another question). The possibility of quasi-agency appears later: after a failure in interaction, either experienced or observed, we might limit or restrain the agency we assign to someone.

9Presumptive agency has a straightforward consequence on trust: it makes trust possible. Here there is no need to argue for a default trust in humans, which is a more intricate issue. Simply, as we expect the humans we encounter to be agents, we are ready to trust them. Agency being a necessary condition of trustworthiness, when we prima facie treat someone as an agent, we treat her as someone who could be trusted. Presumptive agency makes her trustworthiness thinkable.

  • 10 To talk about quasi-agency, it is crucial that this revision of the presumption of agency takes a s (...)
  • 11 Where Q-A is, metaphysically, an agent. The third element, ϕing, can be either a domain of action o (...)
  • 12 The three places are: the one who trusts, the one who is trusted, and the expected action, or the e (...)

10Stable10 deflation of credited agency, i.e. attribution of quasi-agency, can logically take two forms, which are convergent with the two ways in which one can limit reactive attitudes towards a given person (Strawson 2013). The one is three-place: A treats Q-A as a quasi-agent when it comes to ϕing11. This is domain specific, act-relative, circumscribed and situated (“She is not herself when she drinks”, “He is a good researcher but he can’t lecture”). A takes it that Q-A cannot accomplish certain types of actions, or that in given circumstances Q-A is not herself. In this case, A still endorses the participant stance in Q-A’s regards as a whole. A just displays insensitivity to reactive attitudes either in given circumstances (“I don’t take her seriously when she is drunk”) or about certain domains (“Don’t resent him if he’s late, he always is”). A still tends to view Q-A as an agent in most circumstances, but she is ready to see Q-A as a mere quasi-agent when alcohol is on the table or when it comes to teaching in high school. Trustworthiness knows a corresponding limitation: A does not trust Q-A with ϕing / or to ϕ. Three-place quasi-agency determines a negation in three-place trust12. It is clear that Q-A is not taken to be untrustworthy, or suspected, as a whole. She can be trusted, but with an exception. Negations of three-place trust can be understood as negatively specifying the way we relate to someone and not as negatively qualifying the relation as a whole: “I trust her, but not when it comes to alcohol”, or “He is a good researcher [which clearly implies at least some professional trust in him], but I would not trust him to give good lectures”.

  • 13 «Even in the same situation, I must add, they are not altogether exclusive of each other; but they (...)

11The other way to treat someone as a quasi-agent is two-place: Q-A is treated as a quasi-agent. This typically happens when Q-A has serious mental problems or is a child. In Strawson’s words: «the second and more important subgroup of cases allows that the circumstances were normal, but presents the agent as psychologically abnormal – or as morally undeveloped» (Strawson 2013: 344). In this case, A simply steps back from the participant stance, and endorses an objective stance in Q-A’s regards. It is crucial to note that Strawson does not think of the stances as mutually exclusive13. It is rather a matter of measure, and the concept of quasi-agency can preserve this dimension. The less A assigns agency to Q-A, the more Q-A is seen as a quasi-agent and the less A will be ready to endorse reactive attitudes in Q-A’s regards. Two-place quasi-agency determines negations of two-place trust, that is, negations of trust relationships which can ground three-place trust acts (Holton 1994, Domenicucci and Holton 2017). To negate a relation is more complex than to negate an act. Negating three-place trust, means negating a trust act. Negating two-place trust, means negating a trust relationship. Hence, negations of two-place trust do not simply specify the relation, they rather qualify the relation in itself, in the way two-place quasi-agency involves Q-A and the stance taken towards her as a whole. As two-place quasi-agency, two-place lack of trust can take a variety of forms, ranging from mere lack of trust based on inability or incompetence, to active distrust based on motives.

12Whatever the relation we have with Q-A, we can often specify it negatively (i.e. three-place quasi-agency, negation of three-place trust) in a given domain (or for a certain action) without needing to review our whole posture towards her. But this is not always the case. Some things are such that when one is seen as quasi-agent about them, one cannot be said to be trusted in any morally thick sense: “He has no sense of commitment”.

2. The quasi-agential grammar of discrimination

13We introduced a concept of quasi-agency and located it in a Strawsonian landscape. We argued that adult human beings are prima facie credited with a high level of agency, which is necessary to make their trustworthiness thinkable. We stressed that normally they can only come to be treated as quasi-agents in the aftermath of a failure (or series of failures) in interaction. The problem, as it will appear, is that presumption of agency seems to go wrong for the members of stigmatized groups, and this straightforwardly jeopardizes the trustworthiness they are credited with. As it will appear through a quick survey into the grammar of discrimination, identity prejudice determines presumptive quasi-agency for the members of stigmatized categories. In this way, members of discriminated groups are prima facie credited with less individual agency than dominant group members: they are prima facie and systematically treated as quasi-agents. Their being treated as quasi-agents affects their very possibility to be deemed appropriate for trust. Prejudice does not function as an alleged reason for distrust: it rather excludes its victims from the sphere of the reasons for trust.

i) The “quasi-agency test”

14A society officially implementing formal discrimination (caste societies, apartheid societies…) attributes different legal statuses to its members based on specific criteria (gender, race, ascendance…). A disadvantageous legal status limits the discriminatees’ legal agency, i.e. their rights and their freedom. In a modern liberal society that recognizes formal status equality, the type of agency with which discrimination interferes has shifted, from legal to moral agency. Members of discriminated categories see their moral agency deflated, as they are systematically treated as quasi-agents. In an ancient society implementing slavery, the legal apparatus clearly took in charge the formal discrimination. So the legislator, whatever its actual nature, could be held responsible for the limitations imposed on the discriminatees’ legal agency. The mechanism implementing these limitations was explicit, having juridical nature. The situation is less clear in a society where discrimination takes latent and diffused forms. The question of the responsibility for non-legal discrimination is a complex one and we will not address it directly here. The other complex question is how being treated as an inferior agent can affect one’s moral life. For legal discrimination, the causal link between being attributed an inferior legal status and enjoying less power and less freedom was evident. In the case of latent discrimination, it is much harder to unpack how being treated as a quasi-agent contributes to actually reducing one’s power and freedom. In this paper, we stick to easier and more basic questions. We believe that clarifying the mechanism of latent discrimination should be the first step towards both inquiries into the responsibility for discrimination and into the disempowerment of the discriminatees. Therefore, in this section we simply attempt an explicitation of the ways non-formalized discrimination can let the discriminatees be treated as quasi-agents.

15To do this, we do not enter psychological conjectures about the mind or the motivations of the discriminators. We rather focus on the way a society as a whole can produce, register and perpetuate the peculiar moral presumption of quasi-agency for adult members of stigmatized categories. Whether some discriminators are hate driven, others only opportunistic, and some genuinely incapable of seeing outgroup members as agents, is left to empirical psychological inquiry. We are thus moving from moral psychology (1) to social philosophy (2). To translate our account of quasi-agency from the individual to the collective perspective, it is sufficient to rephrase the distinction between agents and quasi-agents in terms of the way in which we talk about them. Let us then introduce descriptions of actions, in a way that allows distinguishing the conduct of an agent and that of a quasi-agent.

16A criterion for agency famously given by Davidson (1971: 7) can help us: «A man is the agent of an act if what he does can be described under an aspect that makes it intentional». Appeal to “descriptions” is crucial here and often overlooked. Interestingly, to grasp this criterion for agency we do not need a view on the mysterious nature of intentionality. We do not need a metaphysical theory to distinguish intentional and non-intentional descriptions of the same action. We perfectly distinguish them for our and others people’s actions in everyday conversation, as excuses and expressions of blame and praise show. A is the agent of event e if and only if there exists a description of e such that e is described as an intentional action of A. Hence, A is not the agent of event e if such description is logically impossible.

17If we only consider the logical possibility of these descriptions, then we can only account for agency as an all-or-nothing property. To make room for degrees in agency (and thus for quasi-agency), we need to step out of the mere logical possibility of these descriptions and take into consideration their social availability. The more intentional descriptions of A ϕing are socially available and frequent, the higher degree of agency is attributed to A in ϕ. We can characterize quasi-agents through this “quasi-agency test”:

Q-A is the quasi-agent of event e if and only if
1) there exists a description of
e such that e is Q-A’s intentional action ϕ and
2) non-intentional descriptions of
e are socially relevant/ widespread/frequent/dominant, in a way that challenges the social availability of intentional descriptions of e.

18The first necessary condition is the logical possibility of e to be described as Q-A’s intentional action. The second is the social frequency condition, which can be formulated positively, in terms of the relevance and dominance of non-intentional descriptions, or negatively, in terms of the way the social availability of intentional descriptions of e is threatened.

ii) Documenting quasi-agency

19The ascription of agency or mere quasi-agency to A is conditioned by the nature of the descriptions of A’s conduct. It is not only the logical possibility of these descriptions, as for the agent/non-agent distinction. It is also a matter of social availability, that is, the frequency and relevance of intentional descriptions. In a society where non-intentional descriptions of A’s actions are dominant, A is treated as a mere quasi-agent.

20The idea that distinctions among presumptive agents and presumptive quasi-agents can be cashed out in terms of the social availability of two types of descriptions of actions, invites us to frame discrimination in a documental perspective, i.e. in light of the theory of documentality for social ontology developed by Maurizio Ferraris (Ferraris 2012). Social reality is understood as being fundamentally a documental reality, that is, a sphere of action that gets its stability through the recording of collective acts, which gives rise, in turn, to social objects. Social entities are recorded collective acts, i.e. interactions registered some way. One’s being treated as an agent or as a quasi-agent should then depend on the way one’s interactions are inscribed in the diffused memory of a society: legal acts, literature, conversations, advertisement, cinema, games, pornography, news, housing policies, fashion.

21A quasi-agent’s behaviour and a fully-fledged agent’s conduct are not socially recorded through the same type of description. The variety of documents, inscriptions and records (Ferraris 2012) that structure a society can be classified on the basis of different criteria. For our purpose, and concerning documents related to a given individual agent, we suggest we should distinguish between records taking an intentional and a non-intentional perspective. Records from an intentional stance tend to focus on individual conduct, framing the agent’s beliefs, motives and desires, whereas records from non-intentional perspectives take the agent’s behaviour as the series of values taken by a bound variable. Individual stories typically make room for personal style, surprise, moral praise and blame, reasons and errors. Records from the non-intentional perspective can be deterministic chronicles just describing factors and outcomes.

  • 14 «A sentence expressing a belief is a bipolar one: it is not enough that we know what in the world w (...)
  • 15 Langton 2012: 140 shows that accommodation to common ground is not just a matter of content and mak (...)

22When insufficiently challenged, the social dominance of non-intentional descriptions about a given set of individuals tends to constitute a set of norms and rules about the socially “appropriate” ways to talk about this group. We can think of it as a grammar of the language that can be used to talk about this group. In this social grammar, non-intentionality appears to be a fundamental rule, one that stays in the background of the language game and that cannot be easily rationally challenged, one that can only be expressed by “framework sentences” (Margalit 1996: 108)14. These framework sentences precisely express the “collective posture” (Margalit 1996) taken toward the discriminatees. Applying Haslanger’s distinction between the two factors jointly constitutive of social structures, schemas and resources, this collective posture clearly appears to be on the side of social schemas (Haslanger 2011: 193-196). Presumptive quasi-agency for the members of stigmatized categories is an «intersubjective pattern of perception, thought and behaviour […] embodied in individuals as a shared cluster of open-ended dispositions to see things a certain way or to respond habitually in particular circumstances» (Haslanger 2011: 194). They encode knowledge, here knowledge about these individuals’ agential power, and provide scripts for interaction, here interaction with quasi-agents15.

23In the next section, we put forward well-known evidence that the conduct of the members of discriminated categories is distinctively recorded as non-intentional behaviour, or sometimes intentional but sub-human behaviour. If we can show that the discriminatees are described through a non-intentional grammar, we can show they are presumptively treated as quasi-agents.

iii) The grammar of discrimination

24Let us now give a hint of some rules in the grammar of discrimination. We sketched this grammar surveying psychological, linguistic and analytic literature on discrimination: it is intended as an open list, and its categories are not mutually exclusive, on the contrary, they clearly overlap. Also, there is nothing in it that is not well known, but we try to set the whole scene. We highlight four rules, four background assumptions that structure socially widespread discourse about members of stigmatized categories:

  1. Essentialist perspective

  2. Deterministic perspective

  3. Objectifying perspective

  4. Sub-human perspective

  • 16 See Dennett: «when one deals with a system – be it a man, machine, or alien creature – by explainin (...)

25Each rule brings its own specific conception of action, its specific theory of behaviour16. Each perspective can be expressed by “framework sentences” (Margalit 1996: 107), i.e. non-bipolar sentences that do not describe facts but give the rules of the way in which we represent these facts.

26The two first rules share the same pattern and have the largest latitude, i.e. they subsume an important variety of descriptions. We can even think of them as jointly constituting the general pattern of the grammar of discrimination, since rules (c) and (d) also share it, and specify it in divergent directions (the sub-human and the objectual models). The linguistic items at the centre of the two first perspectives are generics, open-ended generalizations that omit quantifications. Though we have no figures about that, it is common sense that the number of generics commonly used for outgroup members is significantly higher than for in-group members, given that stereotypes circumscribing the outgroup members are precisely the device allowing generic generalizations. Hence outgroup members are more likely to have their action told or reported through generics, that is, compared to, subsumed under, related to, ascribed to, the purported nature of the group to which they are attributed and the regularities that those belonging to this group are meant to share. Given the frequency of generic use about minority members, these are more likely to see their action attributed to a generic subject, the group to which they are assigned, than to them as individual subjects. Whereas actions of the dominant group members tend to be attributed to individuals, actions of outgroup members are more easily attributed to the group. Hence, consistency, i.e. «the reliable manifestation of trait-relevant behaviours across diverse trait-relevant eliciting conditions» (D’Cruz 2016: 470) is also more easily attributed to the group than to the individual.

  • 17 See Yzerbyt, Rocher and Schadron 1997. Susan A. Gelman’s works about psychological essentialism, an (...)
  • 18 See Rothbart and Taylor 1992: 11-36.

27Let us look closer at the first perspective. This rule can also be understood in light of the relevant evidence from empirical psychology about psychological essentialism in the context of inter-group interactions17, and about the human tendency to conflate social categories and natural kinds18. Ordinary language offers a great variety of grips to essentialist expressions related to gender, race, and other social categories. Just think about the epithet “feminine”, or the genitive “woman’s”. When related to “touch”, “sensitivity” and “beauty”, they ascribe “femininity”, not in relation to any female agent but as denoting the instantiation a self-standing property of “femininity”.

28Generics are especially eloquent and received extensive scrutiny in feminist epistemology. Anderson, Haslangerand Langton, in their 2012 paper, drawing on precedent work by Sarah-Jane Leslie, provide a general framework to understand how generics express and perpetuate discriminatory attitudes. The key element is the double slippage between different sorts of generics, that takes place in their use in ordinary language. Three sorts of generics should be distinguished: “characteristic generics”, “striking property generics” and “majority generics” (Anderson, Haslanger, Langton 2012: 760-767). “Majority generics” have no definitional ambition. You have one of them when a generic resists indefinite singular form. They just denote that the majority of the members of a given group display certain properties. “Striking property generics” are, by themselves, no more definitional than “majority generics”. As opposed to majority generics, they do not imply that the property they ascribe is prevalent in the group. They signal a striking property, i.e. a property “of the sort one would be well served to be forewarned, even if there were only a small chance of encountering it”. Concerning characteristic generics, they “purport to tell what is characteristic of the kind”: they tell what it means to be of a given kind.

  • 19 Donald Trump, in his Presidential bid (16th of June 2015).
  • 20 Cfr: Jones 2013: 197, «Spillover [of distrust] is fostered when the race or ethnicity of minority o (...)

29Both “majority” and “striking property” generics are subject to slippage in the direction of “characteristic generics”, that is, they have a tendency to wrongly essentialize something that is merely either a statistical matter or a salient feature of a minority of specimen of the kind. Trump’s «They [alluding to immigrants from Mexico] are rapists19» precisely plays on this kind of slippage. This sentence ascribes a property, striking by its dangerousness, “being a rapist”, to a social category, (illegal) Mexican immigrants, in a way that, especially through repetition, can become a claim about what it is to be a Mexican immigrant. This kind of slippage, as Anderson et al. (2012) argue, is especially pernicious: these striking property generics turning into characteristics generics are easily accepted as signalling a danger (because of the striking property) and are as resistant to refutation as definitions (under their essentialist interpretation). This linguistic feature should raise awareness on the possible discriminatory effects of mentioning race or ethnicity in crime news when they have no bearing on the crime. Attention is called upon a striking property (“American woman raped by a Latino”). This results in statements like “There are Latinos who are rapists”. This property can then be cast into “striking property generics” like “Latinos are rapists”, and this can result in stable negative perception of Latinos, whose alleged nature seems to include “being rapists”20.

30Analogously, a fundamentally majority generic as “Latinos are lazy” (the example is discussed by Anderson, Haslanger, Langton 2012) easily becomes a characteristic generic defining by laziness the purported nature of Latinos. The non-scientific, instrumental and tendentious interpretation of statistics about labour market easily turns into claims about the nature of Latinos, as the recent Charles Murray case shows21. Noting the threat represented by the slippage from “majority” to “characteristic generics”, should highlight the relevance of forbidden base rates22. That different types of generics are subject to slippage in the direction of “characteristic generics” is a reason to think that the social availability (and sometimes dominance) of generics about the members of stigmatized groups works as a device of essentialization. Taking Dennett’s notion, I suggest that the essentialist perspective encourages endorsing a “design stance”23 when interacting with agents the alleged nature of whom we consider to know (and to be the appropriate object of shared social knowledge). Talking about the discriminatees in generic terms seems to encourage us to view them as specimen of a species, as instantiations of a given nature the essence of which is cast in alleged common sense knowledge (actually, prejudice).

  • 24 Croizet and Leyens 2003.

31The social availability of quasi-intentional descriptions can take a closely related form. Let me call it the “Deterministic perspective”. Here the relevant psychological evidence comes from research on group relations stressing on how causal explanations appear to be more frequently provided for the actions of outgroup members24. Here it is not just that actions of the discriminatees are attributed to a collective subject that ends up being seen as their nature. Here their belonging to a given stigmatized category is taken as a cause or an explanation of their behaviour. Rather than on design, expressions in the deterministic perspective seem to insist on the mechanics behind the scene. Hence we can compare the deterministic perspective to the “mechanical stance”, the other non-intentional stance that we can take toward a system according to by Dennett 1971. Take the Megyn Kelly case, with Donald Trump saying to Donn Lemon, the day after his interview with cnn anchor-woman, «There was blood coming out of her eyes, blood coming out of her…wherever». Here the reduction of Megyn Kelly to less than a human agent is strategic in cutting off her authority, representing her interview as resulting not from rational reflection, but from mere biological determinism: it was not her speaking, but her body.

32As a less extreme example, take the following sentence, addressed by Marc to Paul and referring to their present female friend Jessica : “You know, they can’t help talking”. Formally, this sentence talks about Jessica’s behaviour using a plural, which in this case clearly stands for women, and so somehow refuses to see the specificity of Jessica’s individual conduct. It shifts the focus from her to her alleged category as the ultimate subject of attribution of her acts. Moreover, pragmatically, it has explanatory ambitions about Jessica’s behaviour (it is because she is a woman that she chats, her being a woman determines her conduct). Providing excuses (“Don’t get cross with her: they can’t help talking!”) or justifications (“We can’t go at the movies with her. They can’t help talking”), causal explanations of people’s actions based on their belonging to a social category clearly communicate a deflated conception of their individual agency. Note that in our example the fact that Jessica is present does not seem to refrain Marc from talking about her in the third person. This illocutionary disablement and the causal explanation provided sadly appear as conversationally coherent.

33The essentialist perspective and the deterministic perspective, as, somehow, the “design stance” and the “mechanical stance”, are all the more pernicious that they both respect what Haslanger named the norms of “assumed objectivity” (See Haslanger 1993, Langton 2000 for a presentation, and Langton 1993 for a discussion). Non-intentional patterns of behavioural explanations are actually a central feature of the laboratory view, allowing the isolation of variables and the formulation of predictions. Clearly, under both perspectives, one can only be trusted in an impoverished sense, as in “You can trust the weather”, i.e. as an object of observation and forecast.

34Beside rules of descriptions in non-intentional terms, the grammar of discrimination also has rules that constitute an “objectifying perspective” and an “animalistic perspective”. The first is extensively discussed in both continental and analytic philosophy. From advertisement to pop culture and the fashion industry, there is a variety of social practices that can perpetuate objectification, with their related documents recording and stabilizing an objectifying perspective on their victims. Let us just mention the social artefact that can be viewed as intrinsically recording and circulating objectification, pornography. Rae Langton (1995, 2000 and 2004) developed a detailed account of how objectification works and of what it does, in relation with pornography. The point that is relevant for us is that, as Langton argues about violent pornography: «when men sexually use objects as women, and those objects are pornographic artefacts, whose content is violent or misogynistic, then they will tend to use real women as objects» (Langton 1995: 178). Mainstream pornography, among the other ethical concerns it raises, tends to record and propagate an inert, non-agential image of woman that is a patient rather than an agent. In a situation where “no” is not taken for an answer, trust issues are clearly set aside: objects just cannot fall in the scope of trust.

35An extension of the concept of objectification was put forward by Fricker to explain the distinctive moral harm of epistemic injustice. Being merely treated as a source of information rather than as an informant has clearly something in common with being objectified. However, as Pohlhaus 2014: 102 clearly puts it, «while the idea that testimonial injustice treats an epistemic agent as less than a full epistemic subject seems correct, there is something odd about characterizing this mistreatment as treating someone like an object». Victims of epistemic injustice, who are not recognized as reliable informants, are somehow treated as “partial subjects” rather than mere objects. Pohlhaus’ appeal to Ann Cahill’s notion of “derivatized subjects” is helpful in specifying what is common to epistemic injustice and sexual objectification: in both cases one is treated as someone whose actions and words only count if they support, and conform to, those of the dominator. In this sense, the objectifying perspective is part of the grammar of quasi-agency ascription.

  • 25 Cf : Leyens 2001: 395-411, Leyens 2000: 186-197, Paladino 2002: 105-117.
  • 26 Leyens 2001: 395-411.

36I called the last rule the “animalistic perspective”, as this last pattern of representation of the discriminatees’ behaviour is based on sub-human mental ascriptions. It is the only perspective that makes room for a form of intentionality, but only a sub-human form of intentionality. Psychological research on infra-humanization of out-group members seems to support the idea that it can be difficult to attribute distinctively human mental states to out-group members25 and that there is a differential attribution of uniquely human emotions to ingroups and outgroups26. We are less inclined to attribute secondary emotions, i.e. emotions deemed to be distinctively human, to out-group members. The terrible irony of one of the Party’s mottos in 1984, “Animals and proles are free” (Chapter VII) clearly sums up this reduction of the discriminatees’ agency (here the proles’) to animal behaviour. Here the Megyn Kelly case is again relevant: she is represented as merely reacting to physical stimuli rather than acting as humans are thought to do.

  • 27 «An expression of this is the fact that in many languages the term for a male slave is ‘boy’ (‘puer (...)

37This rule is more insidious than the “objectifying perspective”. It needs not treat persons as straightforward objects, and it needs not either always animalize them, but can infra-humanize with more subtlety, namely through infantilization. An example, analysed by Margalit, is the way the lexicon of slavery is filled with terms usually referring to children, with relative cross-linguistic, cross-cultural and diachronic stability27. Margalit’s example, based on Paul Veyne’s analyses, is the depiction of slaves in Plautus’ comedies. Slaves were represented as displaying agency but of a different kind than human adults’. It was not animal-like agency but child-like: «On the one hand the slaves were seen under psychological predicates, but on the other hand these were predicates appropriate only for the children. Veyne mentions that when Plautus wanted to amuse his audience he described a slave in love. Ascribing full human feeling to slaves seemed as grotesque to his listeners as a complex, passionate love story, taking place in a Kindergarten might strike us. (…) It means treating them like children who will never grow up and become responsible for their actions» (Margalit 1996: 111). Slaves act, to a certain extent (although they are not civic legal agents of course), but they are described as “inherently immature and thus incapable of becoming adults”. Again, the possibility of their trustworthiness is straightforwardly excluded: they are made to be managed, domesticated, or hedged against.

Animalistic descriptions

Non-intentional descriptions

Objectifying descriptions

Type of process substituted for their actions

Biological instinct

Natural phenomenon Regularities


Subject of attribution of their actions

Their body

Group agent Essence

None/ «derivatized subjects» (Cahill)

Type of document

Animal-like and child-like mentalizing

Generics Framework sentences

Pornography Advertisement

Type of description

Intentional but sub-human

Essentialist Deterministic

Non agential

Related psychological evidence

Differential attribution of 2dary emotions to outgroups

Social categories seen as natural kinds Causal explanation of outgroup behaviour

Reduction of empathy after pornographic contents consumption28

Type of posture

Biological stance Infantilization

«Design stance» (Dennett) «Mechanical stance» (Dennett)

«Objectifying stance» (Langton)

«laboratory view» (Haslanger)

Type of «trustworthiness»

Mere reliability based on observation and predictions, Impoverished sense of trust.


  • 28 See the seminal and influential work by psychologist Dolf Zillmann.

38As we briefly saw in sketching these rules, each perspective has its linguistic topical expressions, its related empirical psychological tendency, its own way to see the discriminatees’ actions, and to attribute them to an ultimate “subject” in a way that deprives the discriminatees’ of their acts. Each of these rules is more or less relevant and frequent from one society to another. They are four perspectives from which the discriminatees are looked at.28Despite the variety of these perspectives, they all converge in refusing to see the discriminatees as the ultimate subjects of their actions, the origin of their behaviour, i.e. as proper agents. Agency is denied to a different extent depending on the pattern followed: being treated as an animal, a child, a natural phenomenon or an objet. But in each case one is treated as less than a fully-fledged agent. They are not seen as the origin of their acts. Their conduct is described as dependent on a causal chain that does not have its origin in their human agency, but in supposed instinct, attributed nature, alleged regularities, imposed passivity. In a context where these rules of description dominate a society’s records, the grammar of discrimination constitutes the documental basis for the discriminatees’ deficit of trust.

39It is patent that to be deemed less trustworthy than dominant group members is part of being discriminated. Here I tried to highlight the specific way in which discriminatees are refused trust, introducing the idea that they are denied a necessary condition of trust. We can fail to trust the trustworthy (Jones 2013) in a variety of ways. When it comes to the members of stigmatized categories, this happens because they are refused a necessary condition of trust: presumptive agency. Out-group members’ trustworthiness is not simply dismissed for the wrong reasons, such as identity prejudice. It is rather the reasons for and against their trustworthiness which are silenced. Their trustworthiness becomes fundamentally unthinkable, in the sense that it is straightforwardly taken out of the realm of reasons. The grammar of the reasons to trust, which cannot be but an agential grammar, cannot easily be applied to the socially widespread descriptions of the discriminatees’ behaviour. The very possibility and conceivability of their trustworthiness is made unthinkable. This is done to a different extent, and with more or less domain-specificity, for each group. But as a whole, when trusting them is made socially unconceivable, they are pushed out of social competition. They cannot be considered as average players in the social game, and are more or less cast as enemies. Cooperation with them is excluded, and domination seems the only possibility.

  • 29 See Govier 1993 for an account of the ways in which suffering distrust undermines one’s self-trust.

40Putting trust deficits at the centre of an account of discrimination has other advantages I have not explored here. More specifically, it helps explaining its self-enforcing effects. As Horsburgh 1960: 351 noted, «distrust has other effects which are conditional upon his knowing that [one] is distrusted. The most important of these is moral discouragement». Lack of self-trust29, stereotype threat, and troubles in self-esteem and self-respect could be shown to be bound up with presumptive quasi-agency.

  • 30 Compare with Margalit (1996: 107-109) about responsibility for non decent collective postures: «Peo (...)
  • 31 See Rae Langton’s current research on accommodating versus blocking injustice.
  • 32 I am grateful to Pauline Boyer, René de Nicolay, Pierre Lauret and Huw Oliver for helpful comments (...)

41To end with, framing latent discrimination in terms of a failure to trust the trustworthy has the further advantage to put pressure on the way we ordinarily manage our trust in others. Since it is neither a belief nor an emotion, but the main component of a practical stance, trust can be directly and indirectly controlled (Holton 1994 and forthcoming) and influenced30. This encourages us to pay particular attention when describing out-group members’ actions and trustworthiness, and it has implications in terms of moral education that await elucidation. We are all responsible for blocking31 or spreading quasi-agency presumption and systematic trust deficit32.

Torna su


Anderson, L., Haslanger, S. and Langton, R.
– 2012, Language and race, in G. Russell and D. Graff Fara (eds), Routledge Companion to the Philosophy of Language, London, Routledge: 753-767.

Anscombe, G.E.M.
– 1979,
What is it to believe someone?, in C.F. Delaney, Rationality and Religious belief, Notre Dame (Ind), University of Notre-Dame Press.

Congdon, M.
– 2017,
What’s wrong with epistemic injustice? Harm, vice, objectification, Mmisrecognition, in Routledge Handbook to Epistemic Injustice, London, Routledge.

Croizet, J.-C. and Leyens, J.-P.
– 2003, Mauvaises réputations. Réalités et enjeux de la stigmatisation sociale, Paris, Armand Colin.

D’Cruz, J.
– 2015,
Trust, trustworthiness, and the moral consequence of consistency, “Journal of the American Philosophical Association”, I : 467-484.

Davidson, D.
– 1971,
Agency, in Binkley R., Bronaugh R. and Marras A. (eds), Agent, Action and Reason, Oxford, Basil Blackwell: 3-25.

Dennett, D.
– 1971,
Intentional systems, “The Journal of Philosophy”, 68: 87-106.

Domenicucci, J. and Holton, R.
– 2017,
Trust as a two-place relation, in Faulkner P. and Simpson T. (eds), The Philosophy of Trust, Oxford, Oxford University Press

Ferraris, M.
– 2012,
Documentality. Why It Is Necessary To Leave Traces, New York, Fordham University Press.
– 2007,
La fidanzata automatica, Milano, Bompiani.

Fricker, M.
– 2007,
Power and the Ethics of Knowing, Oxford, Clarendon Press.

Gell, A.
– 1998,
Art and Agency. An Anthropological Theory, Oxford, Oxford University Press

Govier, T.
– 1993,
Self-trust, autonomy, and self-esteem, “Hypatia”, 8, 1: 99-120

Hardin, R.
– 2004a,
Terrorism and group-generalized distrust, in R. Hardin (ed.), Distrust, New York, Russel Sage Foundation.
– 2004b,
Distrust: Manifestations and management, in R. Hardin (ed.), Distrust, New York, Russel Sage Foundation.

Haslanger, S.
– 1993,
On being objective and being objectified, in L.M. Antony and C. Witt (eds), A Mind of One’s Own. Feminist Essays on Reason and Objectivity, Boulder, San Francisco, Oxford, Westview Press: 209-253.
– 2011,
Ideology, generics, and common ground, in C. Witt (ed.), Feminism and Metaphysics, Dordrecht, Springer: 179-208.
– 2014,
Studying while black: Trust, opportunity, and disrespect, “Du Bois Review: Social Science Research on Race”, 11, 1: 109-136.

Hawley, K.
– 2014,
Partiality and prejudice in trusting, “Synthese”, 191: 2029.

Hollis, M.
– 1998,
Trust Within Reason, Cambridge, Cambridge University Press.

Horsburgh, H.J.N.
– 1960,
The ethics of trust, “The Philosophical Quarterly”, 10, 41: 343-354.

Jones, K.
– 2013,
Distrusting the trustworthy, in Archard et al. (eds), Reading Onora O’Neill, Routledge: 186-198.

Kramer, R.
– 2004,
Collective paranoia: Distrust between social groups, in R. Hardin (ed.) Distrust, Russel Sage Foundation.

Langton, R.
– 1993,
Beyond a pragmatic critique of reason, “Australasian Journal of Philosophy”, 71, 4: 364-384.
– 1995,
Sexual solipsism, “Philosophical Topics”, 23, 2: 181-219.
– 2000,
Feminism in epistemology: Exclusion and objectification, in M. Fricker and J. Hornsby (eds), The Cambridge Companion to Feminism in Philosophy, Cambridge, Cambridge University Press: 127-145.
– 2004,
Projection and objectification, in B. Leiter (ed.), The Future for Philosophy, Oxford, Clarendon Press: 285-303.
– 2009,
Sexual Solipsism: Philosophical Essays on Pornography and Objectification, Oxford, Oxford University Press.
– 2012,
Beyond belief: Pragmatics in hate speech and pornography, in I. Maitra and M. K. McGowan (eds), Speech and Harm: Controversies over Free Speech, Oxford, Oxford University Press: 127-147.

Leyens, J.-Ph. et al.
– 2000,
The emotional side of prejudice: The attribution of secondary emotions to ingroups and outgroups, “Personality and Social Psychology Review”, 4: 186-197.
– 2001,
Psychological essentialism and the differential attribution of uniquely human emotions to ingroups and outgroups, “European Journal of Social Psychology”, 31: 395-411.

Paladino, M.P. et al.
– 2002, Differential association of uniquely and non uniquely human emotions with the ingroup and the outgroup, “Group Processes & Intergroup Relations”, 5: 105-117.

Pohlhaus, G.
– 2014,
Discerning the primary epistemic harm in cases of testimonial injustice, “Social Epistemology”, 28, 2: 99-114.

Rothbart, M. and Taylor, M.
– 1992,
Category label and social reality: Do we view categories as natural kinds?, in G. Semin and K. Fiedler (eds), Language, interaction and social cognition, London, Sage, 11-36.

Smith, S.S.
– 2010,
Race and trust, “Annual Review of Sociology”, 36: 453-475.

Strawson, P.F.
– 2013 (1962),
Freedom and resentment, in Shafer-Landau R. (ed.), Ethical Theory: An Anthology, Oxford, John Wiley & Sons: 340-352.

Tetlock, P.E.
– 2000,
The psychology of the unthinkable, “Journal of Personality and Social Psychology”, 78, 5: 853-870.

Yzerbyt, V., Rocher, S., and Schadron, G.
– 1997,
Stereotypes as explanations: a subjective essentialistic view of group perception, in R. Spears et al. (eds), The social psychology of stereotyping and group life, Oxford, Blackwell: 20-50.

Torna su


1 For a review of the sociological evidence, see Smith 2010: 453-475.

2 Jones 2013, focusing on situations where we fail to trust the trustworthy, and Hawley 2014, focusing on partiality in trust, both insist that the relations between trust and discrimination should be of concern for a theory of trust, but they do not provide a general framework of the articulation between discrimination and distrust. Kramer 2004 deals with more specific situations (collective paranoia in intergroup relations) also involving discriminatory distrust. Hardin 2004a also discusses discriminatory distrust, but specifically in relation to terrorist threats. Thanks to Rae Langton for the reference of Jones 2013.

3 By agent, we mean «one who or that which acts; as opposed to the patient, or the instrument […], the efficient cause […], any natural force or substance that produces phenomena» (Shorter Oxford English Dictionary). Humans (as persons, as role-bearers, as representatives…), institutions (a nation, the National Health Service, a currency, the European Union, a firm, a church…), divinities, anything that can be anthropomorphised or have imaginary powers attributed to, can be the appropriate objects of trust and distrust. On the contrary, “I trust that P” is an instance of credence rather than trust, and, in this case, “to trust” is used as a synonym of “to believe”. Intuitively, closely related expectations can be expressed either in the terms of trust – “I trust him to be on time” – or in the terms of belief – “I trust that he will be on time”. This depends on whether the proposition stresses on the agency at work (through the infinitival component for example) or on the expected outcome (here, a that clause describing a state of affairs). The first example, focusing on agency, tends to stress more on normative expectations -expectations of-, as trust tends to do, than the second which is clearly belief-like and sounds more like a forecast-expectations that- (cfr: Hollis 1998).

4 A limit case where she actually produces effects but no one notices, as they affect nothing and nobody at all, seems remote.

5 Note that the notion of quasi-agent is not coextensive with Ann Cahill’s «derivatized subject». As Pohlhaus (2014: 107) puts it, «treating another as one whose subject capacities exist solely in support of and never in tension with my own» makes one a derivatized subject. Hence, all derivatized subjects are quasi-agents, but most quasi-agents are not derivatized subjects.

6 Gell 1998 develops an agential account of art and Ferraris 2007: 193-202 talks about artworks as « artefacts pretending to be persons » (our translation), thus attributing them an especially high level of agency.

7 See Dennett (1971: 104), on a parallel line, about incorrigibility for persons: «For intentional systems that can communicate – persons for instance- the tolerance takes the form of the convention that a man is incorrigible, or a special authority about his own beliefs».

8 I take it that, when one has a gradual concept of agency as I do, responsibility and autonomy imply a high level of agency. Their fully-fledged form is not compatible with quasi-agency.

9 Cfr: Dennett (1971: 93) goes further on this line: «One fact so obvious that it is easily overlooked is that our « common-sense » explanations and predictions of the behaviour of both men and animals are intentional».

10 To talk about quasi-agency, it is crucial that this revision of the presumption of agency takes a stable, durative, dispositional form. Mistakes and excuses, on the contrary, only count as punctual suspensions of agency, not as reductions to quasi-agency, since quasi-agency is a dispositional concept whereas excuses are obviously event-relative.

11 Where Q-A is, metaphysically, an agent. The third element, ϕing, can be either a domain of action or specific circumstances of actions.

12 The three places are: the one who trusts, the one who is trusted, and the expected action, or the entrusted object, or the situated domain of interaction.

13 «Even in the same situation, I must add, they are not altogether exclusive of each other; but they are, profoundly, opposed to each other» (Strawson 2013: 344).

14 «A sentence expressing a belief is a bipolar one: it is not enough that we know what in the world would make the sentence true, we must also know what it would mean for the sentence to be false. A framework sentence such as “She has a soul” is not bipolar. We do not know what it would mean for it to be false. The other human beings have souls –that is, are subjects of psychological predicates- is not a hypothesis but the provision of a framework for representing human beings as such. Framework sentences represent the way to represent our objects» (Margalit 1996: 108).

15 Langton 2012: 140 shows that accommodation to common ground is not just a matter of content and makes the case for a notion of attitudinal accommodation. Rae Langton’s current work in this direction (namely her Locke Lectures) could show more precisely how the conversational common ground of a society’s language games can result in a stable set of moral attitudes (what I call here a “collective posture”) taken at the discriminatees.

16 See Dennett: «when one deals with a system – be it a man, machine, or alien creature – by explaining and predicting its behaviour, by citing is beliefs and desire, one has what might be called “a theory of behaviour”» (1971: 198).

17 See Yzerbyt, Rocher and Schadron 1997. Susan A. Gelman’s works about psychological essentialism, and her discussion of psychological essentialism about gender also special relevance for this point.

18 See Rothbart and Taylor 1992: 11-36.

19 Donald Trump, in his Presidential bid (16th of June 2015).

20 Cfr: Jones 2013: 197, «Spillover [of distrust] is fostered when the race or ethnicity of minority offenders is mentioned even though it has no bearing on the crime».


22 See Hawley 2014: 22 where issues about partiality in trust are articulated with the problem of forbidden base rates, that is, «statistical generalizations that devoted Bayesians would not hesitate to enter into their probability calculations but that deeply offend a religious or political community. […] The primary obstacle to use the putatively relevant base rate is not cognitive but moral», Tetlock 2000: 854 quoted by Hawley 2014.

23 Dennett 1971: 191.

24 Croizet and Leyens 2003.

25 Cf : Leyens 2001: 395-411, Leyens 2000: 186-197, Paladino 2002: 105-117.

26 Leyens 2001: 395-411.

27 «An expression of this is the fact that in many languages the term for a male slave is ‘boy’ (‘puer’ in latin, ‘na’ar’ in the Hebrew of the Bible, ‘boy’ in the South)». Margalit 1996: 111.

28 See the seminal and influential work by psychologist Dolf Zillmann.

29 See Govier 1993 for an account of the ways in which suffering distrust undermines one’s self-trust.

30 Compare with Margalit (1996: 107-109) about responsibility for non decent collective postures: «People cannot directly control what they see. They can do so indirectly. […] Holding framework sentences is an attitude that is not the result of decision. This does not mean that adhering to framework sentences is an unchangeable posture, but that changes in posture do not take place as a result of decision».

31 See Rae Langton’s current research on accommodating versus blocking injustice.

32 I am grateful to Pauline Boyer, René de Nicolay, Pierre Lauret and Huw Oliver for helpful comments and to Maurizio Ferraris, Richard Holton, Rae Langton, Milad Doueihi, Fabienne Peter, Daphnée Setondji and Vera Tripodi for precious discussions on this topic.

Torna su

Per citare questo articolo

Notizia bibliografica

Jacopo Domenicucci, « Trust, Agency and Discrimination », Rivista di estetica, 64 | 2017, 83-102.

Notizia bibliografica digitale

Jacopo Domenicucci, « Trust, Agency and Discrimination », Rivista di estetica [Online], 64 | 2017, online dal 01 aprile 2017, consultato il 19 agosto 2017. URL : ; DOI : 10.4000/estetica.2067

Torna su

Diritti d'autore

Licenza Creative Commons
Rivista di Estetica è distribuita con Licenza Creative Commons Attribuzione - Non commerciale - Non opere derivate 4.0 Internazionale.

Torna su
  • Logo Rosenberg & Sellier