What Attentional Moral Perception Cannot Do but Emotions Can

Abstract

Jonna Vance and Preston Werner argue that humans’ mechanisms of perceptual attention tend to be sensitive to morally relevant properties. They dub this tendency “Attentional Moral Perception” (AMP) and argue that it can play all the explanatory roles that some theorists have hoped moral perception can play. In this article, I argue that, although AMP can indeed play some important explanatory roles, there are certain crucial things that AMP cannot do. Firstly, many theorists appeal to moral perception to explain how moral knowledge is possible. I argue that AMP cannot put an agent in a position to acquire moral knowledge unless it is supplemented with some other capacity for becoming aware of moral properties. Secondly, theorists appeal to moral perception to explain “moral conversions”, i.e., cases in which an experience leads an agent to form a moral belief that conflicts with her pre-existing moral beliefs. I argue that AMP cannot explain this either. Due to these shortcomings, theorists should turn to emotions for a powerful and psychologically realistic account of virtuous agents’ sensitivity to the moral landscape.

1. Introduction

What is “moral perception”? The term can be understood in different senses, and theorists are not always careful to specify which sense they have in mind. So, a better question is this: which, if any, of the mental capacities we actually possess can play the epistemic roles that theorists have hoped moral perception can play? In an important article that recently appeared in the Journal of Moral Philosophy, Jonna Vance and Preston Werner [1] use this approach to advocate the following way of cashing out moral perception:

Attentional Moral Perception (AMP): Humans’ mechanisms of perceptual attention tend to be sensitive to morally relevant properties. Consequently, humans tend to attend to the morally relevant properties they perceive1.

Vance and Werner present evidence which, they claim, shows that “most adult human beings” exhibit AMP [1] (p. 518). Furthermore, they argue that this tendency can explain the moral-epistemic abilities that some theorists have hoped moral perception can account for: AMP can explain virtuous agents’ ability to notice morally relevant features that others overlook; it can explain such agents’ ability to accurately represent properties such as suffering, which are morally relevant; and it can explain their ability to carve out certain features as belonging to more-or-less unified “situations” that are candidates for moral assessment. On this basis, Vance and Werner contend that AMP is the psychologically real phenomenon that plays “all the explanatory roles” [1] (p. 501) ascribed to moral perception by some theorists.

In what follows, I argue that although Vance and Werner are right about what AMP can do, there are certain crucial things that AMP cannot do, things which render it incapable of occupying the central place in moral epistemology that many theorists have ascribed to moral perception. Firstly, many theorists appeal to moral perception to explain how moral knowledge is possible [2,3,4,5]. I will argue that AMP cannot put an agent in a position to acquire moral knowledge unless it is supplemented with some other capacity for becoming aware of moral properties. Second, theorists appeal to moral perception to explain how “moral conversions” are possible, i.e., how experiences can lead agents to form moral beliefs which conflict with their pre-existing moral beliefs [2,3,6]. I will argue that AMP cannot explain why an agent would revise her deepest moral beliefs. This means that, while AMP is surely part of the psychology of a fully competent moral agent, it is far from the whole story—it cannot provide an explanation of how agents are able to sense what is right and what is wrong. In light of these shortcomings, I will argue that theorists should move beyond our perceptual systems; they must look to emotion for a powerful and psychologically realistic account of skilled agents’ perception-like sensitivity to the moral landscape.

The article is structured as follows. Section 2 explains the moral-epistemic roles that AMP can play. Section 3 makes the case that AMP cannot account for moral knowledge or conversions. Section 4 argues that emotions can play the moral-epistemic roles which AMP cannot and that they provide a deeper explanation for AMP itself.

2. What Attentional Moral Perception Can Do

Vance and Werner’s starting point is the need to clarify philosophers’ talk of “moral perception”. Focusing on theorists who claim that virtuous agents have special perceptual abilities vis-à-vis morality2, they differentiate two models of what moral perception might be. The first is this:

Contentful Moral Perception (CMP): “A virtuous agent can represent moral properties as part of the content of her perceptual experience”.

[1] (p. 502)

CMP cashes out moral perception along the lines of the kinds of “high-level” perception advocated by some philosophers of mind [7,8,9]. The idea is that our perceptual experiences present us with a wider range of properties than familiar ones such as shape, colour, loudness, and pitch, which are uncontroversially perceptible. For a skilled forester, such-and-such a tree can appear to be a Scots pine; the phenomenal character of her visual experience presents the tree as being a Scots pine. Similarly, according to proponents of CMP, a skilled moral agent can see or hear an action as cruel, in the very literal sense that her perceptual experience presents the action as exhibiting the property of being cruel.

Although it is advocated by a number of philosophers [10,11,12,13,14], CMP remains highly controversial. For this reason, Vance and Werner go looking for a more “lightweight” [1] (p. 506) model of moral perception, one which can nevertheless provide “all of the explanatory resources” [1] (p. 502) needed to elucidate the moral sensitivity of virtuous agents. To reiterate, the model they settle on is this:

AMP: Humans’ mechanisms of perceptual attention tend to be sensitive to morally relevant properties. Consequently, humans tend to attend to the morally relevant properties they perceive.

AMP avoids the controversial claim that moral properties such as wrongness or cruelty can show up in perceptual experiences. Instead, Vance and Werner rely on the claim that “moral difference makers” can be “subject to attentional focus in perceptual experience” [1] (p. 507). By “moral difference-makers”, Vance and Werner mean the properties that most theorists call “morally relevant properties”, i.e., not the moral properties themselves but the properties that make a given action wrong or cruel (or right or kind, etc.). Standardly, these will be nonmoral properties present in the situation, such as “someone’s wincing in pain” [1] (p. 509) in a situation where someone’s action is wrong partly due to causing said pain3. Thus, while CMP entails that agents can be perceptually aware of moral properties, AMP requires only that agents can perceptually attend to the nonmoral properties on which those moral properties depend. Vance and Werner wisely remain neutral about which nonmoral properties are the morally relevant ones; whatever the correct account is of what makes an action wrong or makes a person cruel, etc., whether that is the agent’s intentions, the action’s impact on people’s wellbeing, etc., the claim made by AMP is that agents will tend to have their perceptual attention drawn to those properties (insofar as they are accessible to perception).

What moral-epistemic work can Vance and Werner’s “lightweight” model of moral perception do? They point to three competencies. The first competency, which they dub Saliency, is skilled moral agents’ ability to notice morally relevant properties that others overlook. Vance and Werner illustrate Saliency with a case put forward by Lawrence Blum in which one agent “is distinctly aware” that a woman in a subway car is uncomfortable, while another agent “is not particularly paying attention to the woman” [16] (pp. 31–32). The difference between the two agents is stipulated to be a difference in their patterns of attention—in what perceptible features of the situation are “salient” to each of them [16] (p. 32). Thus, it is clear that this competency can be explained by AMP. And this competency surely has downstream effects on moral cognition and decision-making, because noticing the morally relevant properties is a crucial first step towards making an appropriate moral judgment.

The second moral-epistemic competency which Vance and Werner claim can be explained by AMP is that virtuous agents tend to have “an increased accuracy […] with respect to identifying the morally relevant features of a situation” [1] (p. 505). They call this competency Accuracy and illustrate it as follows: “It is not just that virtuous agents are more likely to, for example, notice the suffering of the woman on the train. They are also less likely to be mistaken about instances of suffering than the less virtuous agent” [1] (p. 505). Thus, Accuracy means that virtuous agents’ perceptions and/or judgments about properties such as suffering—properties that are morally relevant—are more often veridical than those of less virtuous agents. Vance and Werner’s case that Accuracy can be explained by AMP is fairly indirect: they point to observed correlations between attention to morally relevant properties and morally appropriate behaviour, hinting that more accurate representations of morally relevant properties might be causing more appropriate moral judgments, which in turn cause the appropriate behaviour [1] (pp. 519–520). However, Saliency alone seems enough to explain this putative correlation. Skilled agents’ superior tendency to notice morally relevant properties would explain their superior moral conduct, even if they do not have a superior ability to perceive or judge those features accurately.

Granting for argument’s sake that skilled agents really do exhibit superior Accuracy in their perceptions and/or judgments about properties like suffering, let me offer a more direct argument that AMP can explain this competency. One of the most robust findings in the psychology of attention is that people show greater “response accuracy” for stimuli they attend to than for stimuli they do not attend to [17] (pp. 16–26). In the “spatial cueing paradigm”, arrows are displayed to direct the test-subjects’ attention to a place on the screen where a target is about to appear, or else to distract their attention away to a different part of the screen. Unsurprisingly, subjects report the target’s visual features more accurately when their attention is drawn towards the target rather than to a different part of the screen. This finding has proved so robust that psychologists use response accuracy as a proxy for attention; they routinely assume that a boost in accuracy with respect to certain features is evidence that the subjects are attending to those features. We can infer that agents who attend to the morally relevant properties they perceive will show increased accuracy in their perceptions and judgments about those properties. For example, someone whose attention is drawn to the mental state of the woman on the subway is less likely to be mistaken about whether she is suffering or not. AMP can thus explain Accuracy. And discriminating morally relevant properties accurately is evidently an important precursor to making an appropriate moral judgment, so this is another way that AMP can have positive effects on subsequent moral cognition and decision-making.

The third moral-epistemic competency that Vance and Werner claim can be explained by AMP is one they call Framing. This is a skilled agent’s ability “to identify a situation as one of moral import in the first place, prior to the application of some moral theory” [1] (p. 504). Attention is an inherently selective phenomenon; to attend to certain features is to ignore others [17] (pp. 12–14). Thus, it is plausible that AMP, by restricting the agent’s focus to the morally relevant features at hand, enables agents to carve out certain parts of the scene as forming a more or less unitary situation that is a candidate for moral evaluation. As Peggy DesAutels has argued, “the framing of moral situations” is a crucial first step in “daily moral cognition” which theorists often overlook [18] (p. 337)4. Vance and Werner make a good case that this competency is the work of AMP.

3. What Attentional Moral Perception Cannot Do

As Vance and Werner argue, AMP can explain the moral-epistemic competencies of Saliency, Accuracy, and Framing. However, I will now argue that there are certain crucial things that AMP cannot do: it cannot explain moral knowledge or moral conversions. These shortcomings mean that, while AMP is surely part of the psychology of a fully competent moral agent, it cannot play the central role in moral epistemology that theorists have hoped moral perception can play.

3.1. Moral Knowledge

AMP enables virtuous agents to notice and accurately represent the morally relevant properties of situations they encounter when those properties are perceptible. However, the competencies provided by AMP do not extend to an ability to make accurate moral assessments of those situations. One key limitation lies in the fact that, often, a substantial amount of morally relevant information lies beyond the reach of perception. For instance, in order to judge accurately the moral status of a remark that incorporates a racially charged dog whistle, one needs to draw on background knowledge about social relations and evolving linguistic trends. It is plausible that the moral status of many everyday actions depends on aspects of social and historical contexts that cannot be objects of “attentional focus in perceptual experience” [1] (p. 507), due to being temporally remote, spatially dispersed, or otherwise abstract5. Consequently, in many cases, AMP will not furnish an agent with all the information she needs to make an adequate moral judgment.

An even more serious challenge stems from the epistemological gap separating the nonmoral and moral domains. It is commonly acknowledged that there is an irreducible difference at the level of sense between nonmoral and moral concepts [19,20]. As a result of this, it is doubtful that there are any valid inferences from exclusively nonmoral propositions to moral ones, besides a few trivial cases [21,22]. Now, on the face of it, AMP can only provide awareness of nonmoral facts and properties. If proponents of AMP wish to claim that it provides awareness of moral facts and properties, they need to supply some further account of how it does so. It follows that an agent needs some other means of becoming aware of moral properties in order for the deliverances of attentional moral perception to aid her in making a judgment about the moral status of the situation she is in. AMP thus provides no explanation of how agents can make moral judgments, let alone achieve moral knowledge, unless it is supplemented with some other capacity for becoming aware of moral properties.

A major reason why philosophers have been interested in moral perception is the hope that it can stop the regress of moral knowledge, providing us with noninferential knowledge of moral propositions which can serve as premises for further moral inferences. In effect, many theorists have hoped that moral perception can play the kind of moral-epistemic role that theorists like G. E. Moore [23] and W. D. Ross [24] ascribe to rational intuition, but without facing the charge of appealing to mysterious psychological capacities [6,11]6. AMP, however, cannot live up to this hope. Since AMP provides only awareness of morally relevant properties, it cannot give us an adequate answer to the central question in moral epistemology: how, if at all, is moral knowledge possible? Theorists looking for an answer to this question need to look beyond AMP.

3.2. Moral Conversions

Let us turn to the next shortcoming. AMP enables agents to gather some of the morally relevant information that is needed to apply their existing moral beliefs to new cases they encounter. Thus, AMP readily explains how experiences can lead agents to form new moral beliefs that cohere with their existing moral beliefs. But there are cases in which experiences lead agents to revise their moral beliefs instead. Following David McNaughton [2] (p. 102), let us call such cases “moral conversions”. We find a compelling conversion case in Leo Tolstoy’s autobiographical story, A Confession [26]. Tolstoy describes how his ethical outlook changed when he witnessed an execution in 19th century Paris. As a young man, Tolstoy had believed in an ethics of “progress”, according to which, capital punishment is sometimes warranted as a means of bringing about “evolution” in society [26] (pp. 7–12). Later, he describes how “the sight of an execution revealed to [him] the precariousness of [his] superstition in progress”:

When I saw the heads being separated from the bodies and heard them thump, one after the next, into the box I understood, and not just with my intellect but with my whole being, that no theories of […] progress could justify this crime. I realized that even if every single person since the day of creation had, according to whatever theory, found this necessary I knew that it was unnecessary and wrong, and therefore that judgments on what is good and necessary must not be based on […] progress, but on the instincts of my own soul.

[26] (pp. 12–13)

By his own account, this experience leads Tolstoy to form a new moral belief at odds with his pre-existing moral beliefs. It is not that the experience enables Tolstoy to fill in the details of formerly abstract moral beliefs; rather, the experience leads him to abandon his former moral beliefs altogether.

The problem is that AMP cannot explain such changes in moral view. Arguably, AMP explains why Tolstoy experiences the sensory qualities of the execution so vividly. But it is not as if these facts per se clash with the moral theory he antecedently believes. After all, it is not as if the young Tolstoy was unaware that guillotines separate heads from bodies, or that falling heads make a thump. If one of Tolstoy’s peers had pointed out these nonmoral facts in a discussion of the ethics of progress, they would not have been making a convincing objection. Consequently, if witnessing the execution involved nothing more than attending to these nonmoral facts, we should expect to find Tolstoy drawing out the implications of his existing moral beliefs for the case in hand and continuing to think that capital punishment is justified when it has a net-positive effect on human progress. But that is not what happens: Tolstoy forms a new moral belief that clashes with his previous moral beliefs, and he loses faith in the ethics of progress. We thus face pressure to conclude that something more than attention to nonmoral properties is afoot. In particular, we face pressure to appeal to some state with moral content—a contentful moral perception or, as I propose below, an emotion—to explain his change in view [27] (pp. 586–588)7.

A key motivation for focusing on perception-like sensitivity to the moral landscape is the hope of explaining just such cases. Theorists invoke such sensitivity to explain how moral inquiry can legitimately involve more than just bringing one’s beliefs into coherence8, or how moral deliberation can legitimately involve more than just figuring out how to achieve one’s pre-existing goals [3] (p. 102, p. 107). Since AMP cannot explain how agents can form moral beliefs that conflict with their pre-existing ones, it falls short in this respect too.

4. From Perception to Emotion

Vance and Werner claim that AMP “can play all of the explanatory roles” that some theorists have hoped moral perception can play9. But we have now seen that there are two central moral-epistemic phenomena—moral knowledge and moral conversions—which AMP cannot account for. Where should theorists turn for a more adequate account of virtuous agents’ sensitivity to the moral landscape? In this section, I will argue that the best option is for theorists to turn away from humans’ perceptual systems and look to emotion.

But first, let us briefly reconsider the alternative account of moral perception we set aside in Section 2, namely Contentful Moral Perception (CMP). According to CMP, moral properties such as wrongness or cruelty can feature in the content of our perceptual experiences; a skilled agent can literally see an action’s wrongness in the same way that, some philosophers claim, a skilled forester can see a tree’s being a Scots pine.

I think Vance and Werner are wise to seek an alternative to CMP, because of the lack of supporting evidence for this model. Philosophers of mind who ascribe other kinds of high-level content to perception do so either on the basis of empirical evidence, such as perceptual adaptation effects [35,36,37], or on the basis of phenomenological considerations, such as phenomenal contrast arguments [7,8,9]. Proponents of CMP have found no empirical evidence to support the existence of perceptual experiences with moral content [38]. In earlier work, Werner [12] offered a phenomenal contrast argument in support of CMP. However, by his own lights, his argument fails to rule out the possibility that affective states rather than perceptual content are responsible for the putative phenomenological difference between experiencing something to be wrong and only experiencing it as having nonmoral properties [27,39]10. Given the current state of the debate11, there is insufficient reason to believe that CMP is a genuine psychological phenomenon12.

To take stock, AMP (the “lightweight” version of moral perception) cannot explain moral knowledge or moral conversions; there is no evidence that CMP (the “heavyweight” version) is a genuine psychological capacity. Where should theorists turn for an account of agents’ perception-like sensitivity to the moral landscape?

Well, one psychological capacity that humans undeniably possess is emotional sensitivity to morally significant situations. By this, I mean the capacity to respond to those situations with emotions such as guilt, anger, gratitude, pride, and contempt. An agent’s emotional response to a given object (action, agent, situation, etc.) is driven by morally relevant information about that object—information from perception—but also from memory, thought, imagination, and background knowledge [43] (p. 5). Emotions are thus potentially responsive to a wider range of morally relevant properties than the perceptible ones on which Vance and Werner focus. (This means that emotions avoid the problem that some morally relevant properties are not available to perception, raised in Section 3.1.) The agent’s response to all this information depends on the process of habituation she has undergone in the course of childhood and later life. Some agents become entrained in patterns of emotional response which, from the vantage point of the moral theorist, seem inappropriate: experiencing negative moral emotions towards people who have harmless but unusual sexual preferences, for example. But other agents are brought up to have emotional dispositions that approximate the genuine patterns of nonmoral-to-moral determination which constitute the moral landscape: responding to the right- or wrong-making properties of objects by experiencing fitting13 positive or negative moral emotions towards those objects. This latter kind of agent gives us an alternative way of cashing out virtuous agents’ sensitivity to the moral landscape:

Emotional Sensitivity: Virtuous agents tend to respond to morally significant objects with emotions that fit their moral status.

This understanding of virtuous agents’ sensitivity is in line with historical views such as those of Aristotle [44], Mengzi [45], and Francis Hutcheson [46], as well as some more recent work [47,48,49].

Given certain plausible assumptions, Emotional Sensitivity can explain how moral knowledge is possible. Several theorists have argued that emotions provide defeasible justification for moral beliefs, due to having evaluative content [50,51,52]. On this view, when you experience anger towards x, x seems like an offence to you, which gives you defeasible noninferential justification for believing that x is an offence. More ambitiously, a number of theorists have argued that, when all goes well, noninferential moral beliefs based on emotions can amount to knowledge. Christine Tappolet [53] (p. 173) suggests that emotion-based moral beliefs that are true, and for which there are no defeaters, constitute moral knowledge. In my own work [54], I have argued that, for an agent who is suitably attentive to defeaters, the habit of basing moral beliefs on emotions can be a reliable one. This holds even if the agent’s moral emotions are themselves somewhat unreliable, so long as most of her unfitting emotions are accompanied by defeaters. For an agent like this, the habit of basing moral beliefs on emotions is reliable, because her responsiveness to defeaters effectively filters out most of the unfitting emotions she experiences, meaning she forms few false emotion-based beliefs. Furthermore, if we assume a virtue-reliabilist account of knowledge, this is enough to establish that many of such an agent’s emotion-based moral beliefs amount to noninferential moral knowledge14.

Emotional Sensitivity can also explain moral conversions. It is well known to philosophers of emotion that our emotions can conflict with our pre-existing moral beliefs, a phenomenon known as recalcitrant emotion [58]. This is compellingly illustrated by the case of Huckleberry Finn, who feels insuperably guilty about the prospect of turning in his friend Jim, a runaway enslaved man, despite sincerely believing that he is morally obliged to turn Jim in [59] (pp. 123–125). The same phenomenon allows us to explain Tolstoy’s moral conversion. Initially, Tolstoy sincerely believes that capital punishment is sometimes morally justified. However, witnessing an execution first-hand leads him to experience extreme negative moral emotions such as horror and repugnance. On the basis of these emotions, Tolstoy forms the corresponding moral beliefs that this punishment is grotesque and inherently wrong. In order to restore coherence among his moral beliefs, he revises his previous belief that capital punishment is sometimes justified. In this way, emotions can drive moral conversions [27].

Unlike AMP, Emotional Sensitivity can therefore play the central moral-epistemic roles that theorists have hoped for from an account of moral perception: it can explain how skilled moral agents can discern what is right and what is wrong, even in cases where this conflicts with their pre-existing moral beliefs. But what about the further competencies of Saliency, Accuracy, and Framing, on which Vance and Werner focus? As I have said, I agree with Vance and Werner that AMP is part of the psychology of a fully competent moral agent and that this explains Saliency, Accuracy, and Framing. But I want to close by suggesting that AMP is itself a downstream effect of Emotional Sensitivity.

It is commonplace in the psychology and philosophy of emotion that emotional episodes cause changes in patterns of attention. Among various other attentional effects, the emotion’s object temporarily “captures” the agent’s attention, generating increased focus on the target object while its morally relevant properties are registered (or other evaluatively relevant properties in the case of nonmoral emotions) [41]. This means that an initial emotional response15 directs attentional resources in a manner that yields a more complete and accurate representation of the object’s morally relevant properties. The trajectory of the unfolding emotional episode is then modulated in response to this information, in a recurrent process of emotional reappraisal, with attentional focus held on the object until a stable appraisal is reached [60]. In this way, an agent’s incipient anger might fizzle out once she notices the wry smile on her interlocutor’s face, which reveals that the anger’s target was a piece of well-intentioned mockery rather than a genuine insult. Or an agent’s initial feeling of disquiet might grow into profound sadness as she takes in the scale of the human and animal suffering a wildfire has wrought. Of course, there is no guarantee that a given agent’s emotional and attentional dispositions will lead her to attend to features that are genuinely morally relevant, as opposed to erroneously focusing on irrelevancies while overlooking things that really matter. But a virtuous agent, who tends to experience fitting emotions, will tend to have her attention drawn to the morally relevant properties as her emotional episodes unfold (as argued by Terrence Cuneo [61] (pp. 80–82)).

Admittedly, different emotions bring with them different attentional profiles after this initial phase of attentional capture. In fear, the agent’s attention remains fixated on the object with a view to monitoring how the threat develops, whereas in happiness, the agent’s attention widens to take in other potential sources of value [62]. But theorists tend to agree that these diverse modulations of emotion (which differ by emotion-type) are preceded by an initial stage of attentional capture, which is common to all emotion-types16.

If Emotional Sensitivity brings with it a tendency to attend to the morally relevant properties of the objects the agent encounters, then AMP is explained by Emotional Sensitivity. It follows that Emotional Sensitivity also explains the moral-epistemic competencies of Saliency, Accuracy, and Framing that AMP brings with it. Thus, as well as explaining how moral knowledge and moral conversions are possible, Emotional Sensitivity provides a deeper explanation of all the abilities discussed by Vance and Werner17. The best explanation of Saliency, Accuracy, and Framing thus traces back to emotions.

5. Conclusions

Vance and Werner [1] set out to identify a “genuine psychological phenomenon” that can “play all of the explanatory roles” that some theorists have ascribed to moral perception. I hope to have shown that the phenomenon they point to—Attentional Moral Perception—can indeed play some of the explanatory roles associated with moral perception but that it cannot explain how moral knowledge or moral conversions are possible. Attentional Moral Perception is thus incapable of taking the central place in moral epistemology that many theorists have ascribed to moral perception. However, I have argued that a different psychological capacity—virtuous agents’ tendency to experience fitting emotions towards morally significant objects—provides a satisfying explanation of moral knowledge, of moral conversions, and of Attentional Moral Perception itself. To reach a satisfying understanding of skilled agents’ perception-like sensitivity to the moral landscape, theorists ought to move beyond our perceptual systems and look to emotions18,19.