2.1. Rules and dual-process theory

Hare is a total act-utilitarian,Footnote ² but typical human agents are more likely to maximize aggregate utility (the well-being of all actual and possible beings), not by regularly relying on explicit consequentialist thinking, but by employing some hybrid decision procedure. According to TLU, that procedure consists in “prima facie principles” (Hare Reference Hare1981: 38), or rules, that we should internalize such that violating them, or considering doing so, causes in us feelings of guilt and compunction (ibid. 29–31). In other words, when we have internalized the correct rules we are disposed not only to act in certain non-utilitarian ways but also to experience feelings that express our commitment to such rules.

Following Varner, call the rules of intuitive-level system ILS rules. The substantive thesis that we should often rely on simple moral heuristics follows from the empirical thesis that we are cognitively limited. So, TLU can both explain our intuitive reactions to standard utilitarian cases and justify some of those reactions in ordinary contexts – on the ground that having such reactions tends to promote utility. Hare provides two stylized character descriptions of the two extremes between which moral thinking takes place. The “archangel” is not subject to normal cognitive limitations; they can think exclusively critically. They would know, at any given time, which of the options available to them would maximize utility. In contrast, the “prole” has cognitive limitations that are severe enough that they would do best to think exclusively intuitively (Hare Reference Hare1981: 44–46). Most of us are neither archangels nor proles; “we all share the characteristics of both to limited and varying degrees and at different times” (ibid. 45). Most of our everyday decisions should be made at the intuitive level, but some unfamiliar contexts will call for more deliberation. Critical thinking is deployed to reflect on the rules themselves “in a cool hour” – on whether and how they should be overridden, revised or replaced. For the rules that we internalize can conflict or fail to provide sufficient guidance or need to be updated, especially in novel cases. So, when we have enough information, time and cognitive resources, critical thinking supplements rules to determine the best course of action. For instance, we might be justified in violating the rights of some people to save a large number of lives. Because such circumstances are exceptional, good ILS rules include strong constraints against harming persons. Rules that have a “deontological flavor” can be more likely to do good given what we know about the world and human psychology.

Dual-process theories posit two distinct cognitive systems and/or types of processes. “System 1” (fast, automatic, emotional, stereotypic, subconscious) and “system 2” (slow, effortful, deliberative, rational, conscious) (Kahneman Reference Kahneman2011) roughly correspond to Hare's intuitive and critical levels. Joshua Greene draws a helpful analogy: the two systems are like the “point-and-shoot, automatic” settings and “manual mode” of a digital camera (Reference Greene2013: 132–43). One is used frequently, the other less so. One is more emotional, the other more cognitive and deliberative. System 1 is more efficient and inflexible, and the “central tension” of our dual-process brains is to negotiate trade-offs between efficiency and flexibility. Manual mode thinking, which allows for controlled, conscious reasoning, and which we all share, is predisposed to utilitarian thinking (ibid. 198). And it's meant to solve problems that automatic settings can't solve reliably – problems such as global poverty and climate change that are unfamiliar to the brains our ancestors evolved.

The two levels are thus two sides of a unified psychology that sometimes conflict but can complement each other. In fact, the descriptive part of Hare's theory predicts that our intuitions will resist the verdicts of the critical level, and in most circumstances, this is as it should be. At the same time, dual-process theorists tell us that our intuitions will often misfire, in which case conscious, controlled reasoning should step in. On Varner's interpretation of TLU, ILS rules include rules of personal morality, common morality, professional ethics, and the law. ILS rules work best most of the time because they yield quick and easy answers. General though they are, they can be fine-grained enough to accommodate different individual “temperaments” and “capacities” (which fall under personal morality) as well as the particular circumstances of different societies (e.g., through law and common morality). For different individuals or societies different sets of rules apply. Rules should be general enough to release individuals of the burden of critical thinking, yet specific enough to accommodate particular circumstances.

Varner believes that TLU is uniquely equipped to avert both standardly consequentialist welfarism and anti-utilitarian abolitionism about animal use, by articulating a decision procedure involving non-utilitarian elements and critical utilitarian reasoning. Part of that decision procedure includes different rules for different kinds of animals.

2.2. Categories and moral status

Varner divides animals into persons, near-persons, and merely sentient, and outlines the possible sets of rules that should govern our interactions with them. The three distinct, if fuzzy, tiers reflect a spectrum of cognitive complexity and yield a corresponding three-tier account of moral status.

Merely sentient

Most if not all vertebrates, as well as cephalopods, are most likely to be sentient, i.e., have “the capacity for phenomenally conscious suffering and/or enjoyment” (Varner Reference Varner2012: 108). Most sentient animals are “merely sentient” (i.e., lack the capacities instantiated by the other two tiers) and most animals currently raised and slaughtered for food fall under this category.

Persons

A person is a rational, self-conscious, autonomous being with a biographical sense of self. In order to construct self-narratives, one must have concepts (of self, birth, death, and personality), hence a natural language (chapter 6). Persons can tell life stories, desire to be a certain kind of person, and have a unique interest in how their “lives-as-a-whole” go. This implies understanding oneself as having desires and being the subject of a story pursuing them, but also having desires about one's desires. These capacities give the lives of persons greater moral significance, and their higher moral status is translated into rights or side-constraints embedded in ILS rules. According to Varner, we currently have no good evidence that nonhuman species qualify for personhood.

Near-persons

Some animals sit in between the merely sentient and persons. They have autonoetic awareness, or a sense of self over time (chapter 7). Great apes, dolphins, elephants, scrub jays and other corvids are near-persons; rats and monkeys might be. While near-persons cannot tell stories about themselves, they can engage in forward- and backward-looking thinking. Some can even perform “mental time travel,” that is, anticipate and remember future and past states of the world and their own experiences in context (i.e., episodic memory; Tulving Reference Tulving1985). Some can recognize themselves in mirrors and/or engage in some form of mindreading. Autonoetic awareness allows individuals to re-experience past events and to make future plans that matter to them. Their lives are thus more “morally charged,” containing more opportunities for well-being. These animals can therefore be harmed or benefitted in more ways.Footnote ³

Varner's theory recommends incremental yet significant reforms of animal agriculture. Industrial animal agriculture and fishing are morally impermissible because they fail to respect sentient animals’ most basic interests, and this failure is not justifiable by the human interests that are thereby satisfied. The principle “Don't kill sentient animals unnecessarily” should be incorporated into the ILS rules of all societies (Varner Reference Varner2012: 229), along with a principle that sentient animals deserve respect (in a rather thin sense of direct consideration for their interests). But the principle does not specify exactly when killing animals is permissible or required. An ILS prohibition against unnecessary killing does not apply to practices that are on balance optimific. Indeed, an absolute ban on raising and killing farm animals at the intuitive level would be non-optimific, according to Hare and Varner. To see why, let us consider the replaceability argument, to which both are committed.

4.1. Visions of humane sustainability

Varner lays out three types of visions that each come with different standards of assessment.

Pre-lapsarian visions are the visions of “ancient societies that survived relatively unchanged for millennia,” which Varner calls so “because environmentalists often venerate such societies and assume that their descriptively sustainable practices also met whatever norms are implicit in their conception of sustainability” (Reference Varner2012: 260). Native Americans were once justified in hunting bison in ways that caused great suffering and sometimes prevented them from using all of the animals killed. But modern, industrialized societies may no longer permissibly do so given the available technology and our understanding of animal welfare. Because ILS rules display a “benign form of relativism”, “the drive hunts of pre-contact Native Americans may well have been an example of humane sustainability for persons in their circumstances … [but] surely modern humans would not be justified in using such inefficient and relatively inhumane hunting techniques” (ibid. 262–63). TLU explains why a given practice can be wrong in our circumstances but permissible in others.

Contemporary visions are attempts to improve animal welfare using contemporary science and engineering “constrained, at least generally, by contemporary consumer preferences and economic realities” (Varner Reference Varner2012: 262). Animal welfare reforms are now accessible and scalable for modern husbandry and slaughter practices. Despite these significant improvements, contemporary practices, exemplified by factory farms, run afoul of the basic principles of good ILS rules. Nonetheless, certification programs, new regulations, and changes in attitudes pave the way for the approximation of a vision of humane sustainability.

Utopian visions involve radical change from contemporary visions and motivate greater long-term progress by “shifting the goal posts of common morality,” enabling large-scale shifts in people's perceptions of animals and their moral significance. Utopian visions facilitate the gradual internalization of new rules that better reflect the verdicts of critical thinking. They “give concrete expression to the ideals of utilitarianism” and can make “achievable what once were utopian proposals” (Varner Reference Varner2012: 256). Utopian visions are not implementable in the near term (because the principle of utility has a built-in “conservative bent or inertia”), but future generations will update the ILS rules of their ancestors, and by approximating utopian visions “in their personal moralities,” individuals “can help move the goal posts of common morality and make their implementation in future generations more readily justifiable from a Harean perspective” (ibid. 260).

According to Varner, we should disallow the killing of animals, not even in utopian visions, at least now. Animal activists, he writes, “live a lifestyle that it would be unethical to force upon society at large at present, but which might become feasible as social norms in future generations… . In the long run … [they] can influence entrenched cultural assumptions and, over time, reduce the transition costs of reform” (ibid. 270). For now, “a reasonable case can be made for eating small quantities of meat from extensively reared ruminants (e.g., cattle) as a form of demi-vegetarianism” (ibid. 283). Varner offers an illustration with Ernest Callenbach's (Reference Callenbach1996) “Buffalo Commons” proposal:Footnote ⁶

Bison being merely sentient, Varner finds the proposal “highly appealing aesthetically” and suggests that “bison rearing and slaughter could be made tremendously humane, given their natural behavior and resilient health, and modern carcass-processing technologies” (i.e., in-situ selective slaughter, with minimal pain, stress or herd disruption) (ibid.). Even though bison meat is still is a “niche market,” the proposal is utopian in its “grand scope … a vision of humane sustainability that can be seen as inspiring to contemporary consumers and producers. Just describing it goes some small way toward moving the goal posts of our common morality” (ibid. 275). In fact, Wild Buffalo Ranchers in South Dakota “claim to do just this.”

Varner's framework remains succinct, so let me make some extra assumptions. The difference between contemporary and utopian visions is a matter of time and perspective. The benefits of utopian visions are to be assessed in the long run, in part on the basis of their aspirational effects. Further, visions focus not just on individual behavior but on the factors causing or facilitating behavior. They do so through ILS rules and the social, political, and institutional infrastructures that support them. Finally, the process is mutually reinforcing: new attitudes and behavior support higher-level changes, higher-level changes shape attitudes and behavior, and behavior signals and reinforces attitudes. On these assumptions, a vision aimed at producing good ILS rules should include a timeline and a specification of its targets (legislation, public policy, moral education, consumer base, social media, etc.). I will argue that Varner's utopian visions may inadvertently foster the type of harmful attitudes that they are designed to suppress. In a nutshell: the permissible use of sentient creatures for food sets the bar too low for respect (i.e., promoting animals’ interests for their own sake). As I argue now, Varner's visions miss important features of their target audiences.

The argument has two steps. First, utopian visions require a certain connection to contemporary visions in order to shift the goal posts. I briefly lay out the desiderata of an effective utopia – which provides motivation for greater change rather than entrenching the status quo. Second, because the promotion of visions has opportunity costs, a case for specific visions requires comparisons to alternatives, and I suggest that Varner's visions do not come out ahead on this front.

4.2. Shifting the goal posts

The function of utopian visions of humane sustainability is to inspire contemporary consumers and producers. For various reasons, we can assume that truly humane agriculture would not work well at scale (by its own ethical standards) (Schlottmann and Sebo Reference Schlottmann and Sebo2019: 108–10). The question is what trickle-down effects utopian visions can have on contemporary visions. The key, Varner and I agree, is encouraging a dramatic reduction in consumption of meat and animal products. Can humane agriculture deliver on its promise?

Let's suppose that visions like the Buffalo Commons would bring about significant benefits – by restoring grasslands, increasing biodiversity, enriching soil quality, and enhancing carbon sequestration; by giving pleasant lives to animals that could be harvested humanely with mobile processing plants and efficient slaughter techniques; and by enabling sustainable agricultural practices that benefit farmers and local communities. Using the same land for other purposes, whether it be untouched grasslands, less sustainable extensive ranching, or growing crops, would not produce more marginal utility. Let's concede too that, if they were enforced by archangels, such visions would maximize total happiness. Even so, the comparative benefits of different rules depend on the consequences of internalizing and enforcing them. In other words, Varner must show that associating humane sustainability with the continuation of using animals as sources of food and ecological services at the intuitive level will most likely shift goal posts in the right direction. For utopian visions to be effective, and worth striving for, they must be such that, if enough people were to internalize them, they would actually improve animals’ lives. But, as Varner acknowledges throughout, these are complex empirical issues.

Something the Harean wants to avoid is unintended spillover: for instance, that embedding replaceability in utopian visions could have a widespread negative effect on contemporary visions by entrenching the status quo, where eating animals is seen as morally acceptable. And because contemporary visions involve billions of animals, the stakes are high. Critical thinking, ruled by utility, can act as a safeguard, but it cannot do the work alone. We must ensure that ILS rules will reliably implement the correct verdicts of critical thinking. Under these conditions, why not think that animal liberation, or any relevant approximation, could have the desirable effect on common morality by challenging the status quo?

4.3. Opportunity costs

Humane agriculture, by promoting niche markets (e.g. grass-fed beef, pasture-raised pork, and cage-free eggs), does not involve very many animals. So whatever harms it still causes to animals should not be cause for concern relative to their overall positive effects. And, from replaceability, we can infer that we are doing more good than harm by giving these animals happy lives. Plus, more people become sensitive to animal welfare, and laws and common morality will shift accordingly. Correct?

Not so fast. Promoting utopian visions has opportunity costs. Time spent promoting humane agriculture is not spent promoting plant-based alternatives. So even if humane utopias did not have a negative effect on animals over time, alternatives might still have better consequences. And contrary to what Hare (Reference Hare and Jamieson1999: 241) thought, demi-vegetarians may not have a “very powerful advantage over full vegetarians when it comes to influencing the market.” Farmers are rewarded by converting primary resources and labor into animal protein efficiently. Unless it is economically beneficial for them to do so, they have no incentives to take animals’ interest directly into account. Welfare standards are primarily a response to demand from well-off consumers and constrained by economic interests more than moral considerations. A Harean might reply that humane sustainability would put constraints on economic trade-offs since animal welfare is a fundamental desideratum of humane sustainability. But the initial question recurs: can demi-vegetarianism and humane agriculture, as utopian visions, deliver on their promise of shifting the goal posts more effectively than alternatives?

First consider what it would take for alternatives to gain traction. Meat is so inexpensive because the production and distribution costs are kept low by economies of scale, efficiencies, government support (subsidies, checkoffs, and buybacks), and unregulated cost externalization (Bollard Reference Bollard2019). One way to make alternatives competitive is to increase their consumer base and for new interest groups to organize. Producing scalable, sustainable plant-based alternatives could re-anchor groups to new social norms much more effectively than humane agriculture. Goal post-shifting may involve disrupting the global food system rather than promoting small-scale animal husbandry.Footnote ⁷

Still, we can treat what visions would be most effective at inspiring widespread change as an open question, especially if we remember that utopian visions involve a timeline and moving targets. We should distinguish between advocating and adopting visions. Generally, campaign A might cause more people to decrease their consumption of animal products, and by greater margins, than Campaign B, independently of whether we should actually adopt the diet recommended by A or B. Thus, Varner cites “certain statistics” suggesting “that Hare may be right,” that people are not ready to be converted to full vegetarianism (Reference Varner2012: 272). Yet Varner acknowledges that “which has greater ‘propaganda value’ – demi-vegetarianism or veganism – is an empirical [question].”

Indeed, but there are some empirical reasons to think Hare may not have been right. Recall that ILS rules must be relatively inflexible to be efficient. So, when we make ethical consumption choices, we must also think about how they will affect our habits, which will in turn affect our future actions, and so on. Schlottmann and Sebo (Reference Schlottmann and Sebo2019: 189) write: “one reason … to take a relatively strict approach to ethical consumption … is that anything less than that makes it difficult to develop the relevant habits, and so to live up to the relevant ideals in the hustle and bustle of everyday life.” To conscientious omnivores and demi-vegetarians, we might say: “if we allow ourselves to eat meat at all, then we will likely end up eating meat more often than we should” (ibid.). This slippery slope argument is behavioral and might turn out to be incorrect, but if we can already personally suspect that we would act rightly less often if we adopted a flexible diet, then we know we should internalize stricter habits. After all, the very reason we need intuitive rules is that the more complexity we introduce in our decision procedures, the less likely we are to deploy those rules reliably. The finer-grained the rule, the more room for error, complacency or rationalization (also see John and Sebo, Reference John, Sebo and Portmoreforthcoming). Indeed, Rothgerber (Reference Rothgerber2015) found evidence that conscientious omnivores are less likely than vegetarians to believe they should strictly adhere to their diet, and partly as a result are more likely to violate their diet and feel less guilt when doing so. One explanation seems to be “difficulty following one's diet.” This might be surprising since conscientious omnivores “have more options than meat abstainers” (ibid. 202), but it makes sense if finer-grained (i.e., more complex rules) are harder to abide by.

Moreover, flexible ethical consumption lends itself to moral licensing, “the tendency to see what one is currently doing as all one needs to be doing in order to be a good person” (Schlottmann and Sebo Reference Schlottmann and Sebo2019: 190). I suggested that demi-vegetarianism may further entrench the status quo of perceiving farm animals as edible. Moral licensing likely contributes to that process too. Demi-vegetarianism looks less demanding than a full commitment to conscientious omnivory, but by the same token it lends itself to complacency, ultimately acting as a drag on moral change.

We must also consider the social context of our choices – our friends, family, colleagues, and the businesses we patronize. We influence them and they influence us. In social contexts, compliance signals approval of a desirable social norm (Bicchieri Reference Bicchieri2017). Legality and market availability also signal social acceptability.Footnote ⁸ So, if most people eat meat, even in small quantities, most people will uptake the norm that eating animals is acceptable (Delon Reference Delon2018; Schlottmann and Sebo Reference Schlottmann and Sebo2019). The smaller the gap between utopian visions and what is relatively easy under contemporary visions but still counts as progress, the smaller the effect of utopian visions. Of course, too great a gap could be paralyzing too, making meaningful progress seem unattainable. Still, this leaves plenty of leeway. Since our behavior is largely a function of social norms, behavioral change requires the norm to shift, which, in turn, requires that enough people be ready to change their behavior for a new norm to emerge (Bicchieri Reference Bicchieri2017). So, we must choose policies that we are disposed to follow and that make us less likely to cause harm. The above empirical speculations suggest that alternative visions like animal liberation may hold more promise for inspiring change.Footnote ⁹

This doesn't mean we cannot meet in the middle. By focusing exclusively on short-term gains, incremental welfare campaigns (e.g., cage-free eggs and Meatless Monday) yield immediate benefits, although they may fail to sustain profound, durable shifts in attitudes. Off-putting, confrontational tactics, direct action, and abolitionist messaging have immediate costs, although the long-term benefits of promoting a society built upon compassion rather than exploitation and killing may outweigh such costs. We would do best to see these different tactics as mutually reinforcing rather than exclusive (Sebo and Singer Reference Sebo, Singer and Gruen2018), as Varner eventually does. He considers Singer's remark that, even if raising animals for food may be justifiable critically, “the level of practical moral principles” (i.e., ILS rules) is different:

Singer's position is actually more nuanced, acknowledging that different audiences will be more receptive to different diets and lifestyles and that “[i]t is probably a good thing if different people and organizations promote each of these goals” (Singer Reference Singer and Jamieson1999: 325; Varner Reference Varner2012: 273). Again, Varner treats this point as an open question. He seems confident that we can draw fine-grained distinctions among animals that are good to eat and those that are not, as he is that we can compartmentalize the intuitive and critical levels (more on this shortly). I'll argue it's not quite clear.

Still, one could contemplate different visions concurrently across different audiences, or across time, moderate visions making people more favorably disposed toward radical ones over time. If these visions are mutually reinforcing rather than exclusive (one utopia being nearer on the timeline than the other), the transition would nonetheless require that we already start promoting the more radical vision, at least to some audiences. I suspect Varner might agree. But this doesn't settle the case yet.

Let's recap first. What utopian visions we should promote (i.e., advocate rather than adopt) depends on a range of empirical considerations, including a timeline and a specification of the target audiences of particular sets of ILS rules.Footnote ¹⁰ But if the above speculations are correct, we should design ILS rules guided by sufficiently radical utopian visions, lest we fail to deliver on the promise of humane sustainability. How exactly to articulate such utopian visions so as not to hinder change in the here and now is an open question, but, at any rate, our end-point should not be humane agriculture.

In the next section, I argue that TLU falls prey to compartmentalization problems.

5.1. Williams’ objection to Hare

I argue that Varner's conception of the value of lives conflicts with a standard understanding of moral status and the cognitive templates that constrain good ILS rules. This is not to say that no agent could possibly be equipped with the relevant psychology, but the point of designing descriptively adequate ILS rules is to accommodate us more or less as we are.

Let's begin by reviewing Bernard Williams’ objection to Hare and Varner's response to the problem – or problems – of compartmentalization. According to Williams, Hare's TLU involves an incoherent moral psychology. It's important for TLU to withstand the objection for the sake of descriptive adequacy. Hare anticipated the problem:

This is what Williams suggests is impossible. Williams criticized Hare in two main instances (cf. Miller Reference Miller2014).

Williams is making two related points. First, dispositions (i.e., prima facie principles, ILS rules) are seen from a certain perspective, which affects their character. And since these dispositions are designed to be effective, they cannot do their job if they are to be seen or experienced from two incompatible perspectives. Second, their effectiveness turns on our intuitive perception of certain things as intrinsically valuable, which, again, the act-utilitarian standpoint undermines. This point, I think, straightforwardly applies to our perception of the value of animal lives.

In the second instance, Williams wrote that

Williams’ critique is internal: ILS rules cannot do their job because they're chosen using critical thinking (from the lens of maximizing aggregate utility). The motivational force of rules depends on their shaping the way we see the practical world. The key step in Williams’ argument is that we cannot see something both as intrinsically valuable and as instrumentally necessary (aggregation-wise). When we see rules as mere instruments, the objection goes, we are not using them in the way they were designed to be used – we can no longer do so – yet this is precisely what TLU implies for critical thinking to play its role. By treating animals as edible commodities, we threaten the resilience of intuitive dispositions by irreversibly introducing market norms in the moral domain (see e.g. Ariely Reference Ariely2008: 76). Williams did not think that it was “impossibly difficult” to recover from the utilitarian perspective, but “that the thoughts are not stable under reflection” (Reference Williams, Seanor and Fotion1988: 190).

Both Hare and Varner feel capable of shifting between the two perspectives, and keeping them distinct as required (Hare Reference Hare1981: 52; Hare Reference Hare, Seanor and Fotion1988: 289–90; Varner Reference Varner2012: 84). Varner adduces some empirical evidence that many people are capable of what Williams thought a true two-level utilitarian would have to possess: not just gut reactions evolved by our ancestors, but dispositions that would have to be both motivationally significant and open to critical inquiry.

The main piece of evidence comes from dual-process theories, but it's unclear whether it supports TLU. The literature does show that people rely on two distinct types of thinking, but our intuitions are not those that would have been designed by critical thinking. If anything, dual-process theorists emphasize the unreliability of “automatic settings,” or system 1. Our intuitions are likely to be unreliable because they are rather inflexible (Greene Reference Greene2013: 217). For instance, they are insensitive to scope and susceptible to “modular myopia”: “our brains have a cognitive subsystem … that monitors our behavioral plans and sounds an emotional alarm bell when we contemplate harming other people… . [T]his alarm system is ‘myopic,’ because it is blind to harmful side effects” (ibid. 224). Our intuitions can be unreliable, even if they work well in many contexts, because “[a]ll automatic settings rely on specific cues that are only imperfectly related to the things they're designed to detect” (ibid. 227).

Here's the worry: TLU seeks to design efficient ILS rules that will, by their very nature, be “rather inflexible.” Given the goal of fostering moral consideration for sentient animals, they will likely have to rely on harm-based cues. But precisely insofar as they are inflexible, there are going to be contexts where our harm-detecting module may clash with the permissibility of eating animals. This is, presumably, where Greene's manual mode would kick in, to override the automatic settings and lay out the utilitarian rationale for replaceability. We are, after all, “emotionally blind” but not “cognitively blind.” So far, so good. But if the literature on cognitive dissonance is any indication, as I discuss shortly, we should expect this process to be quite uncomfortable, which Greene would be the first to concede. The challenge is for that discomfort not to backfire and undermine the rather inflexible moral consideration owed to sentient animals according to ILS rules.

Dale Miller argues that our reactive attitudes are resilient enough to survive access from the objective standpoint. We can't “avoid experiencing [them] in the ordinary way the vast majority of the time. We can displace these attitudes with an objective attitude by choice, as in Hare's cool hour, but not for long” (Reference Miller2014: 54). Likewise, Miller suggests, we can alternate between seemingly incompatible types of moral thinking over time. Williams would have to show that a TLU agent “would always think critically and intuitively simultaneously” (ibid. 55). But, Miller and Varner claim, we can compartmentalize our perspectives to preserve the viability of each system.

I agree that reactive attitudes are resilient. Perhaps Williams was wrong that our intuitive dispositions would not survive utilitarian thinking. Still, the following must be true to dissolve the problem of compartmentalization. First, dual-process theories must be correct – in itself a strong assumptionFootnote ¹² – so that system 2 can override system 1. Second, intuitive thinking must be resilient enough that system 2 does not prevent us from resuming thinking intuitively when we need to. Of course, we should not trust system 1 to tell us when we should do so, and so we're left to wonder how system 2 can lead us to resume thinking intuitively. Yes, critical thinking designs ILS rules, but can it tell us how to shift gears? There must be a sweet spot of rules that are resilient enough to fulfill their role yet flexible enough that critical thinking can control them. This is what a descriptively adequate TLU seems committed to. Respect-cum-replaceability must be among such rules. Section 5.2 will cast doubt on this possibility.

5.2. Compartmentalizing animals

On one influential model, moral cognition consists in the perception of two minds. Roughly, people attribute moral properties on the basis of mind perception. Attributions of moral standing track perceptions of “mindedness,” typically along two dimensions: Experience (e.g. suffering, feelings, emotions) and/or Agency (e.g. thinking, memory, planning, self-control) (Gray et al. Reference Gray, Gray and Wegner2007). Morally significant events involve a “moral dyad,” a pairing of a (blameworthy) agent and a (vulnerable) patient: say, a villain torturing a kitten; a hero rescuing an old lady. The moral dyad, with harm at its core, constitutes the fundamental template for perceiving moral violations (Gray et al. Reference Gray, Young and Waytz2012; Gray et al. Reference Gray, Schein and Ward2014).

A growing psychological literature has established the role of mind perception – and factors and biases that influence it – in our attitudes and behavior toward animals (Goodwin Reference Goodwin2015; Jack and Robbins Reference Jack and Robbins2012; Kasperbauer Reference Kasperbauer2017; Sytsma and Machery Reference Sytsma and Machery2012). The “meat paradox” is the well-documented fact that most people (claim to) believe it is morally wrong to harm sentient creatures yet (believe that it is permissible to) eat animals. As a result, meat-eaters tend to experience cognitive dissonance when holding both attitudes. In order to reduce the discomfort, they have two options: “one can reject meat consumption, bringing one's behaviors into alignment with one's moral ideals, or one can bring one's beliefs and attitudes in line with one's behavior through various psychological maneuvers” (Piazza et al. Reference Piazza, Ruby, Loughnan, Luong, Kulik, Watkins and Seigerman2015: 114). Motivated cognition explains much of it (Piazza and Loughnan Reference Piazza and Loughnan2016). People can deny that animals have the psychological features that ground their moral status, typically the capacity to suffer: Bastian et al. (Reference Bastian, Loughnan, Haslam and Radke2012) indicates a negative relation between the perceived edibility of an animal and how “minded” it is perceived to be. People can also deny that animals deserve moral concern to begin with: Loughnan et al. (Reference Loughnan, Haslam and Bastian2010) primed subjects with either dried beef or dried nuts and found that the former leads to decreased attribution of a mind and moral status to cows. Bratanova et al. (Reference Bratanova, Loughnan and Bastian2011) found that when some animals are categorized as food (e.g., Papua New Guinea's tree kangaroos), participants attribute to them less sentience and lesser moral status. The fact that people engage in motivated cognition in order to reduce dissonance confirms the tendency to attribute a mind, and consequently moral standing, to other animals. A series of studies also found that omnivores are motivated to adopt a range of rationalizations for eating meat, including the so-called “4Ns”: that it is natural, normal, necessary, and nice (i.e., pleasant) (Piazza et al. Reference Piazza, Ruby, Loughnan, Luong, Kulik, Watkins and Seigerman2015).

These facts – the centrality of harm-based concern, the tendency to motivated reasoning – are central to the way we relate to other animals and might undermine the prospects of respect-cum-replaceability, because replaceability leads us to regard animals as deserving less consideration than they do, whereas respect leads us to disapprove of harming animals. And our acceptance of replaceability is all too likely to trigger complacent motivated reasoning. If these facts constrain ordinary psychology, ILS rules should be mindful of them.

Varner is aware of the pitfalls of building too much utilitarian thinking into ILS rules. Recently, he has argued that ILS rules “the internalization of which disposes one not to think like a utilitarian during the moment-to-moment management of many interpersonal relationships” should apply to our relationships with our pets (Varner Reference Varner and Overall2017: 77). Because of the overall benefits of positive relationships with pets, he argues, pet-keepers should not see them as replaceable. For “loving another sentient being requires you not to think of it as ‘replaceable’ the way everyone and everything is when you are doing explicitly utilitarian thinking” (ibid.). The ILS rules governing our relationships with farm animals are notably different. This aligns with our ability to compartmentalize animals between different categories (Amiot et al. Reference Amiot, Lépine and Zaky2019). But, if Varner is correct about pets, the replaceability of farm animals precludes a range of attitudes that foster mutually beneficial relationships with them (of the sort that conscientious omnivores commonly claim to promote). In fact, why not imagine that Varner's (Reference Varner and Benatar2002) conception of “domesticated partnerships” could provide utopian visions for all domesticated animals? Like companion animals, farm animals could generate revenue (through tourism, therapy, veterinarian care, or labor) in non-exploitative relationships that would exercise their faculties. Since Varner assumes that we should not regard our pets as replaceable, let alone eat them, it's unclear why we really need different ILS rules for different categories of animals, especially if the benefits of small-scale humane slaughter-based agriculture are bound to be so restricted by scale.Footnote ¹³

Admittedly, in some sense replaceability is consistent with respect. As many cat and dog lovers will attest to, we regard our companions as worthy of respect, we provide them with individualized care, yet when they die, it's not infrequent for us to “replace” them in some sense. How does this differ from the sense of replaceability implied in the replaceability argument? For one thing, we mourn our deceased companions, remember them qua individuals, and don't so much try to replace them as we seek other companions to fill their role. They are never truly replaced, which is why we grieve. On the other hand, animal farming treats animals as strictly replaceable: the primary reason for bringing new beings into existence is that they, in turn, will be replaced by other animals in the context of market exchanges. Such interactions are closed once the animal dies, and parties are accordingly substitutable. This is what market norms allow for, through impersonal relationships (Maguire and Brown Reference Maguire and Brown2019). Markets would cease to be as efficient if farmers developed the sort of non-substitutable relationships that genuine respect requires. Market efficiency is a major benefit of impersonal transactions (ibid. 10–13). In this sense, replaceability is hardly compatible with respect, if respect implies a complex of personal, open-ended attitudes and dispositions.

Couldn't ranchers internalize certain ILS rules such that they would want to care for bison for their own sake, not just for profit? They would have them humanely slaughtered while sincerely believing that death does not violate any morally significant interests, since the merely sentient have no interest in continued existence. They would be respecting bison while treating them as replaceable. This is plausible, and remember that ILS rules can be tailored to variations in “personal morality.” Yet I doubt that this can be generalized at the intuitive level – especially among consumers, who do not stand in any meaningful relationships with farm animals and treat them, as a matter of fact, as commodities. If it turns out that more people than I suspect can in fact respect animals while regarding them as replaceable, then this particular compartmentalization problem dissipates. I would welcome a two-level utilitarian argument to this effect. But Varner himself doubts this is feasible with pets; it remains unclear why this would be much easier with farm animals. We haven't fully solved, I conclude, the problems of compartmentalization.