Brandom and Ethics

Jon Cogburn has just put up a post about the ethical implications of Brandom’s thought (here). As much as I respect Jon, I’m afraid I almost entirely disagree with the post. I think he’s being really unfair to Brandom. I mean no offence to him, but his claim that some of Brandom’s remarks (to the effect that pain has no intrinsic moral significance) are evil strikes me as hyperbole. I haven’t yet fully gone through Reason in Philosophy, but I’ve been thinking about the ethical implications of Brandom’s work (see my speculative heresy piece on ethics) and have come to very much the same position expressed in these remarks, yet without any of the more horrific implications Jon seems to see in them. To warn you, this is another fairly long post (coming up 7,000 words).

1. Explicating the Problem

First of all, we should not be damning Brandom for ways his philosophy could be misunderstood or used to justify things that neither genuinely follow from it, nor that he himself assents to. We shouldn’t blame Nietzsche (or Kant) for the horrors of Nazism, and we should treat Brandom in the same way. This being said, Jon’s worry is that Brandom’s philosophy somehow legitimates the mistreatment of creatures that lack the rational capacities necessary to engage us in conversation, and thus to count as sapient (rather than merely sentient) on by Brandom’s standards. Examples of these are pre-linguistic children, humans that cannot develop language skills or have lost them through some means (e.g., brain damage), and non-human animals. I think that there is something to this worry, and that we do need to show why this is not the case. Doing this properly would require developing a more robust ethical theory out of Brandom’s approach. At minimum, we simply need to show that this is possible, and hint at what it would look like.

However, I think that Jon sets the bar against which Brandom is to be measured far too high. The reason he gives for taking Brandom’s remarks to be evil is that they imply that the sentient creatures discussed above have no intrinsic moral worth. We should bare in mind that denying such creatures intrinsic moral worth is far from equivalent from legitimating their mistreatment. Now, some might say that this claim could nonetheless be used to justify such mistreatment. Yet, that someone could attempt to so justify atrocity by appeal to this idea does not for that matter show that it actually follows from Brandom’s work. We should keep in mind the above principal and not hang Brandom either for the motivation that misunderstandings of his work could provide, or for justification that a genuine understanding of his work could hypothetically (although it has not been shown to) provide. The real problem here is precisely what something like intrinsic moral worth is supposed to amount to.

Now, lets look at the Brandom passage Jon quotes:-

Here the question is: should moral respect go with sapience or with biology? Sentience-utilitarians such as Singer say the latter because they see morality as normatively driven by the intrinsic sensuous evaluation implicit in the phenomena of pain and pleasure. But we kantians see the normative basis of morality as derived form the positive freedom of giving and asking for reasons. Mammalian sensuousness, sentience, is at best a necessary condition of that, not a sufficient one. According to this line of thought, it is the capacity to engage in conceptual activity, being a subject of sapience, not of sentience, that is in the first instance morally significant. This does not, of course, settle it that we should not accord respect and rights to embryos, as potential moral persons, or for that matter, to non-human animals. But in each case the argument appeals to an indirect connection to the primary subjects of moral respect and (so) rights: discursive creatures. Kant certainly thought it was wrong to cause pain to animals for no reason–but that is not in the first instance because of what it does to them, but because of what doing that to them does to us. (Brandom 2009, 148)

I’m afraid I’m much in agreement with what Brandom says here. There are a lot of ethical theorists who try to ground ethics in the intrinsic normative force of pleasure and pain, and who interpret these in biological terms. Setting aside for the moment issues about the nature of normativity, and whether anything like intrinsic normative force makes any sense, such accounts are at the very least insufficient. This is because they tend to limit pain and pleasure to the simplest physical forms of satisfaction, and end up treating things like the satisfaction and frustration of our wishes and projects as subsidiary matters whose significance depends upon the stimulus they produce in us. A desire that my brother go to university is only fulfilled by that state of affairs coming to pass, and not if my family engages in an elaborate deception which convinces me that it has (and thus produces the corresponding pleasure). The disconnect is even more obvious in the case of desires which aim at things one cannot possibly experience (such as that my philosophical contribution is remembered after my death). If we take this view, we can find no good reason not to actively deceive others and ignore the wishes of the dead. Moreover, such approaches exclude the possibility of moral agents (and patients) that have radically different physical structures. At the very least, there needs to be some independent basis for the ethical responsibilities we owe to genuinely rational agents.

Now, Brandom goes further than simply criticising the sentience oriented account, and claims that rationality is the root of the ethical, and that rational agents are the primary subjects of moral responsibility. Jon thinks that Brandom is drawing a false dichotomy here, forcing us to choose between either an account of ethical obligation based on biological sentience or rational sapience. He thinks that this is dichotomy is false because it is possible to have it both ways:-

One can have absolute obligations to not cause needless pain to all sentient creatures, as well as obligations to respect the autonomy of rational creatures. There is absolutely no good reason for Brandom to commit a false dichotomy here. The latter obligation in no way undermines the former.

The question is this: if we can have it both ways, what exactly is the common source of ethical obligation? Indeed, how do we make sense of the idea of normativity on this basis?

It is indeed possible that we can have an absolute moral obligation both to prevent harming sentient creatures and to respect rational autonomy, in the sense that the two obligations aren’t incompatible. However, it is just as possible that we might have an absolute ethical obligation to wear hats on Tuesdays and to respect the paths of lava flows (and indeed, only those obligations). What we really want is to know why such obligations are foundational. The problem is not that the obligation not to harm sentients and the obligation to respect the autonomy of sapients are themselves incompatible, but rather that the accounts of why they are foundational are incompatible. To reject this dichotomy we either have to reject the demand for an explanation of why there are such absolute obligations, or we need to provide a proper account of the twofold root of ethical obligation in sensation and reason. I think the former is undesirable in the extreme, and the latter seems impossible (although I’m willing to be proved wrong).

Given this, I think Brandom’s account of the nature and basis of normativity in terms of rational autonomy is the best such account there is. This holds even if we discount the other successes of his approach in providing a systematic account of understanding, language use, practical action, etc. We have seen above that there are seemingly insurmountable difficulties for the sentience-based approach in accounting for the obligations we owe to rational entities. Yet, it is possible to give an account of the way that pleasure and pain play a role in rational motivation, and thus have a derivative kind of normative force. Principally, this lets us account for why we shouldn’t inflict pain on (or withhold pleasure from) rational agents (unless of course there is good reason to), but it doesn’t yet say anything about whether we should extend this courtesy to non-rational agents.

This is of course where Jon’s worry comes in. Brandom claims that although his Kantian (sapience-based) approach does not on its own settle this issue. He takes it that more needs to be said to determine whether we have derivative (rather than foundational) responsibilities to sentient creatures of various stripes and what they are. However, Jon thinks that Brandom is disingenuous at this point:-

I’m sorry, this is damningly and obviously stupid in so many ways. Yes it does settle the issue. The uncombined atoms that go on to make up the fetus are a “potential person,” yet they clearly have no moral worth. And the idea that non-sapient creatures have rights only because treating them in certain ways effects us can only be the idea that in harming them it somehow causes us to harm sapient creatures (not I suppose by being more likely to cause humans pain, since pain and pleasure have no moral regard, but rather in decreasing other human’s sentience). For Brandom it is wrong to torture a severely retarded person or animal just in case that would cause us to somehow make other humans less sapient. But that is transparently evil as well as psychologically implausible (Arendt’s terrifying banality of evil is just the way that human beings can compartmentalize in ways that completely undermine what Brandom is suggesting). But this means that if the Nazis had stopped at killing retarded people and done it in a way that did not cause undue stress to the non-retarded, there would be absolutely nothing wrong with that. I’m sorry, if such a view isn’t evil then the word has no meaning.

There are several points that must be responded to here. Firstly, assuming that Brandom has a definite account of what ‘potential’ means when he references it as part of an open question is just unfair. Brandom was referencing an existing debate, not taking up any kind of position in it. Demonstrating the vacuousness of simplistic uses of the notion of potentiality has no relevance here, as Brandom is not taking up any concrete position vis-a-vis the nature of potentiality, let alone potential personhood. Secondly, Brandom at best suggested the uptake of something like Kant’s approach to the status of non-sapient creatures. He didn’t provide a fleshed out account of what this would consist in. However, Jon has chosen to take this as an endorsement of the crudest possible form of such an account. The claim that: “For Brandom it is wrong to torture a severely retarded person or animal just in case that would cause us to somehow make other humans less sapient” not only presents Brandom’s position as far more determinate than it is, but the notion of “less sapient” is not even properly defined. We are left to imagine that on Brandom’s account we can legislate as much cruelty as we like as long as we can still have stimulating dinner party conversation about it later. Such a view might well be evil, but it doesn’t seem to be Brandom’s view.

Nonetheless, rejecting such a crude version of this kind of account is not good enough. As noted at the beginning, we need to show the possibility of deriving ethical obligations to non-sapients from within something like a Brandomian account, even if we needn’t work it out in its entirety. I’m going to go out on a limb and offer a possible strategy. It’s not fully worked out, but I think it has some promise.

2. Sentience and Sapience

We first need to tackle the core distinction between sentience and sapience, which Jon seems to suggest is another false dichotomy. I actually have certain problems with the way Brandom uses this distinction, mainly because it tends to suggest that the two terms are of the same kind, like the difference between dogs and cats. However, what one must remember is that sentience is a biological, and thus natural category, whereas sapience, the status of being a rational subject, is actually a form of normative status. Being a rational subject is something one is counted as within the practice of giving and asking for reasons. To be counted as such is to be counted as having certain fundamental rights and responsibilities, which are themselves normative statuses that cannot be cashed out in naturalistic terms.

This doesn’t mean that there isn’t any link between the ascription of sapience and the real causal structure of sapients described in natural terms. The ascription of sapience is based upon an assessment of capacities to engage in the game of giving and asking for reasons. Brandom has given his most detailed exposition of this idea in Between Saying and Doing (although I must admit that I didn’t entirely digest the stuff on AI). This means that there could be naturalistic facts that obligate us to count something as a sapient. It should be noted here that Brandom doesn’t require one to be a logical being to be a sapient being. One needn’t be able to use logical language, or make explicit inferences, only be able to track implicit inferential connections between claims and actions. Although I don’t think Brandom has said this explicitly, I think a good maxim is that anyone we can count as sapient, we should count as sapient. If we recognise that someone has certain capacities, then we are obliged to treat them as an autonomous rational agent. The salient point is that we can give an account of an entity’s capacities in naturalistic terms, and thus it is possible that we could show that certain entities with particular causal structures should be counted as sapients.

However, although there can be naturalistic facts about an entity’s constitution that are sufficient for us to ascribe sapience, it’s not clear that there is a set of such facts that are both necessary and sufficient. There could be entities that have radically different causal structures than us that we would nonetheless be obliged to count as rational agents. The capacities that motivate the ascription of sapience are very much multiply realisable. This contrasts with the notion of sentience as deployed by sentience-utilitarians. It is a biological notion linked to certain particular ways it can be realised. Of course, there are plenty who would say that sentience is not a biological notion, and would appeal to fuzzy definitions of what it is like to have conscious experience. I have expressed my general dislike of such hazy ideas of ‘consciousness’ elsewhere (here). When Brandom says that “Mammalian sensuousness, sentience, is at best a necessary condition of that, not a sufficient one”, the important phrase is ‘at best’.

So, Brandom has no problem accepting naturalistic (and evolutionary) accounts of how it is that we come to have the causal capacities necessary to be counted as fully fledged rational agents, and thus has no problem with such accounts picking out similar capacities we share with animals. His issue is that such an account would never be equivalent to an account of rational sapience itself. For Brandom, the divide between the natural and the normative is inviolable, and I tend to agree with him. Jon mentions teleosemantics as something that could further fill in this gap between the sentient (natural) and the sapient (normative) at one point, which is well known as a program that tries to derive the normative from the natural. Brandom has explicitly argued against teleosemantics (see ‘Intentionality, Modality, and Normativity’) and although I won’t repeat his arguments here, I think they are effective. Jon also mentions the Heideggerian tradition of grounding theoretical understanding in practical understanding. I do think Brandom needs to tell a better story about how our practical grasp of ordinary action (rather than our practical grasp of the game of giving and asking for reasons) contributes to our grasp of conceptual content. Other than BSD, his first paper on Heidegger (‘Heidegger’s Categories in Sein und Zeit‘) is the closest he’s come to doing this, and although I think it’s a misinterpretation of Heidegger, there is some promise there. However, I don’t think these Heideggerian lessons would do anything to ameliorate the split between the normative and the natural in the way teleosemantics does.

Moving on, the thing to take away here is that the moral worth of autonomous rational agents, our obligation to respect their autonomy, is not really a natural (or empirical) matter, even if we are obliged to count certain kinds of natural entities as autonomous rational agents. There is nothing like intrinsic moral worth here, if by this we mean some special metaphysical property possessed by a certain class of entities. Rather, the obligation that we have first to recognise certain entities as rational subjects, and then to respect their autonomy, is a transcendental matter.

This is because there are transcendental norms that we are bound by insofar as we engage in rational discourse or action at all. These norms are transcendental because they provide the condition of the possibility of rationality as such. They dictate what it is to undertake theoretical or practical commitments, by specifying both what one must do to do so, and what one does in doing so. Given that norms are themselves a form of practical commitment, these transcendental norms also form the condition of the possibility of normativity itself. To make these norms explicit is thus to give an account of the nature of normativity in normative terms (which is a virtuous, rather than a vicious circle). There is no non-normative ground of the normative – as McDowell is fond of saying, it’s norms all the way down. I’ve explained these points in more detail elsewhere (hereherehere and here).

The important point is that although most of these norms would not normally be considered ethical (Is it strictly unethical to contradict oneself?) they nonetheless provide the ground of ethical normativity in two senses. In the weak sense, they provide the ground of normativity as such, and thus also of any species of normativity. In the strong sense, they provide us with specific obligations to other rational agents, and these constitute our obligation to respect their autonomy. At this point we have a broadly Kantian ethical story, although it must be noted that Brandom does not endorse the categorical imperative, because he has a more nuanced story about how autonomy functions. I’m not going to go into this in detail, because I haven’t done enough work on it. Suffice it to say that there are a variety of possible positions regarding precisely what is involved in respecting the autonomy of rational agents. Habermasian discourse ethics is the other example that comes to mind.

3. Derivative Intentionality and Dependent Obligations

The question is now how we derive obligations to non-sapient creatures on the basis of this transcendental story. Non-sapients can’t have intrinsic moral worth, but then, there is a good sense in which neither can sapients. This transcendental approach is antithetical to the notion of intrinsic value as such. However, it seems that what is really at stake is not intrinsic value, the denial of which can hardly be considered evil, but rather the independence of our obligations to non-sapients from our obligations to ourselves and other sapient creatures. This is a complicated matter, because there are varying levels of dependence that such obligations could exhibit. At one extreme we could have some form of universal obligation to non-sapients that is completely independent of whatever relationship they stand in to sapients, and at the other we could have purely local, particular obligations, such as those not to interfere with animals that are the property of others sapients. There is a whole spectrum between these two however.

I don’t think its possible to make our obligations to non-sapients completely independent of all relations they stand in to sapients. The transcendental basis for ethics lies in our obligations to other sapients, and there just isn’t any way around this. The question is thus what degree of independence we can out of this. What we want are some general obligations that extend beyond particular relations that non-sapients stand in to particular sapients. This would undercut the crude interpretation which makes the wrongness of harming non-sapients dependent upon the relation of the action to me. I think the key to finding this lies in two features of Brandom’s account: his account of derivative intentionality and his account of authority and responsibility.

Derivative intentionality is Brandom’s account of what we ascribe to non-linguistic creatures (such as animals) when we treat them as having intentional states (such as beliefs and desires), so that we can provide intentional explanations of their behaviour. This intentionality is derivative because our ability to ascribe it is parasitic upon our ability to engage the game of giving and asking for reasons. This is because we are in effect ascribing normative statuses (such as theoretical and practical commitments) to these creatures in the same way that we ascribe to our interlocutors in rational debate. Now, there are other forms of derivative normative status that we can ascribe to entities, such as biological and technological functions. However, the way in which we ascribe them is not parasitic in the same sense that derivative intentionality is.

To explain this, we need to explain the relation between authority and responsibility. This involves explaining a bit more about the normative status of sapience, or being a rational subject. In essence, a subject is a locus of responsibility. A subject is just something that can be held responsible for the commitments that it has undertaken. This is what the identity of a subject consists in. In this respect, a lot of the classical debates about the metaphysics of personal identity have everything upside down – the identity of the subject across time is not the basis on which we hold it responsible for things it has done in the past, but rather, us holding it responsible is primarily what this identity consists in. This isn’t to say that there aren’t specific criteria governing whether we do in fact ascribe identity in this way, just that these don’t pick some special metaphysical substrate of authority and responsibility. In principle, I could be uploaded onto a computer in some fashion, and it would be possible to say that I am the same person, because it makes sense to say that I retain all of the responsibilities I had previously (at least, those that aren’t conditional on me not being uploaded into a computer). However, a subject is just as much a locus of authority. A subject has a unique authority over which theoretical and practical commitments it undertakes (although this is not for that matter an absolute authority). It is also in virtue of this unique authority that I can acquire any other kind of authority, such as military command, or ownership. For me to be the same person is also for me to retain this unique authority.

Now, when we ascribe normative statuses to non-sapient creatures we ascribe to them a normative status analogous to being a rational subject. However, this status is deficient, insofar as such creatures cannot directly exercise any authority over what they are committed to. Now, when we keep track of the normative statuses we ascribe to ourselves and our interlocutors in discourse (what Brandom calls deontic scorekeeping), we do not just keep track of those commitments an individual explicitly acknowledges. On the one hand, we also keep track of the individual’s consequential commitments (those that are implied by those they acknowledge), but on the other we also keep track of implicitly acknowledged commitments. This is to say that we ‘read between the lines’ and try to ascribe to our interlocutor a bunch of normative statuses that best makes sense of their behaviour. This kind of thing is familiar from Davidsonian radical interpretation. The ascription of derivative intentionality consists entirely of this ‘reading between the lines’. This is what makes the ascription of derivative intentionality different from the ascription of functions.

What makes this possible are genuine similarities between non-sapients and sapients. It’s important to note that this process of ‘reading between the lines’ is not a purely theoretical activity, but is something that depends on a practical grasp of the behaviour of humans that is partly acquired and partly inbuilt. Prediction is a two way street, and our practical ability to predict others’ behaviour feeds our ability to ascribe them intentional states, just as the latter enables us to make predictions of their behaviour on the basis of intentional explanations. The important point is that the similarities we share with non-sapients tap into this practical grasp to differing extents. This is not to say that it isn’t possible to ascribe derivative intentionality in a theoretical fashion, such as taking MRIs of a monkey’s brain and using the similarities in causal structure to project intentional descriptions. Both similarities in external behaviour and internal structure potentially count. However, the extent of the similarities limit richness of the kinds of intentional interpretation we can achieve. This is fairly apparent in the different relationships we have with dogs and lizards. It is a lot harder to ascribe to a lizard a rich mental life without pushing into fiction. A final, and crucial point to make here, is that pleasure and pain are absolutely fundamental similarities in this regard, they play a big role in letting us fill the background in terms of which we make any sense of a creatures behaviour, in virtue of the systematic role they tend to play in rational motivation.

4. Quasi-Agency and Structural Norms

In looking for general obligations that we have to non-sapient sentients, I think we have to further examine the normative status that we confer on non-sapients when we ascribe derivative intentionality to them. For want of a better term, lets call this quasi-agency, as opposed to rational agency. To do this it is necessary to think a little bit more about the nature of normative status.

Normative statuses are socially instituted, and they are generally instituted within the context of a certain practice that is governed by norms. The example I usually use is that of being a goal keeper in football (or soccer), which confers on one a certain role within the practice, consisting of specific obligations and permissions. However, as we have explained normative statuses can be ascribed to non-humans. A knight in a chess match has a various normative statuses, its status as a knight, which indicates how it may be moved in general, and its status as occupying a current position, which delimits how it may be moved now. A ticket to a concert has a normative status that it conveys on its holder, namely, permission to enter the concert. These are both kinds of functional status. One might detect a little whiff of Heidegger here, and one wouldn’t be wrong. The other kinds of status we have so far discussed are rational agency, which is the status of being counted as an interlocutor in the game of giving and asking for reasons, and the various subsidiary normative statuses that such agents can take on, such as theoretical and practical commitments, and the corresponding entitlement of such commitments. These latter statuses are special, in that they get there significance from the transcendental norms governing the practice of giving and asking for reasons. The other kinds of status discussed are based on contingent norms that are in some sense instituted by a social group.

However, I think that there is a third kind of norm, which we might call a structural norm, that is neither transcendental nor entirely contingent in relation to them. To explain this it is important to recognise the distinction between norms and practices that Brandom endorses, and that I’ve tried to elaborate on (here). Norms are a form of collective practical commitment, and are thus themselves normative statuses. Practices are collective behavioural regularities, which are usually self-sustaining to some extent, and can be fully described in natural terms. Norms are implicit within practices, such that we can engage in a process of interpretation (which itself is a matter of giving and asking for reasons) that makes the norm explicit in the form of a rule or principle. What needs to be pointed out is that although the norms governing the process of giving and asking for reasons are transcendental, the practices to which they correspond are as natural as any other. The biological and social factors that enable us to engage in the game of giving and asking for reasons have evolved naturally and must perpetually sustain themselves, just like any other. The question is, how do they so sustain themselves?

I think the answer is that they are interwoven with a whole bunch of subsidiary structural practices that make possible there current state and continued perpetuation. Implicit in these practices are structural norms. The best example of this I can think of is child-rearing. It takes a decent amount of time and attention to raise a child not only into a language user, but into a genuinely autonomous rational agent. What’s interesting about this is not only that the two aren’t the same, but that there isn’t a crisp boundary delimiting either. A baby has no authority, but neither does it have any responsibility for its actions. As it grows and starts genuinely acting we begin interpreting its behaviour, ascribing it intentions, and we also impart it with a kind of limited responsibility, in terms of which we try to correct its actions (through scolding and the like). As it develops linguistic competence our need to ‘read between the lines’ lessons, which means that we slowly confer on it a form of authority over its commitments.

We think of these kinds of provisional authority and responsibility as sandbox authority and responsibility, and they constitute a varying scale of quasi-agency. The practices of parenting (which include the wider social activities in which children are brought up), although they are derivative upon the game of giving and asking for reasons, are not specified by the transcendental norms of rationality. They are nonetheless an essential part of the wider context of the rational way of life. As such, the normative status of quasi-agency we confer upon children gets its significance from structural features of our rational form of life. We thus seem to have located a complicated set of obligations that have we to non-sapients (or even proto-sapients), that are not dependent upon particular sapients, but which rather rest on a wider social state of affairs.

5. Conclusion: The Halo Effect

This basic approach requires some more explanation, and some further consideration about how it can be extended to other cases. I think the obvious counter-example that is going to be raised against this kind of approach are those of infants who can never develop language, let alone get out of the forms of sandbox authority we set up for them. From here it is only a short distance to cases of permanent and temporary mental illness and debilitation. I think these cases can be accommodated, but it involves further understanding the difference between norms and practices, and the way practices tie into the hardwired mechanisms that enable our practical grasp of one another’s behaviour.

To reiterate, the basic idea is that although there cannot be obligations to non-sapients that are entirely independent of some relation to sapients, there can be some general obligations to non-sapients (or types of non-sapients) that are independent of any particular sapients, but are dependent upon certain structural practices tied to sapience (or at least, to the way we realise sapience). The essence of the idea is that the obligations get their force from the indispensability of the corresponding practices for maintaining and propagating the wider practices constitutive for sapience. The important thing to note here is that this indispensability is actually a causal matter. These obligations emerge not from the transcendental norms of rationality, but from the causal role that the practices in which they are implicit play in maintaining the practices constitutive for rationality. For this strategy to work, we need to posit something like a modified version of the categorical imperative, i.e., an obligation not to any particular rational interlocutor, nor even to all rational interlocutors, but to the very institution of rationality itself. This needs to be worked out in more detail, but if true, it would lend a transcendental force to otherwise contingently instituted structural norms.

The question remains as to how we extend this strategy beyond the case of genuinely proto-sapient children to non-sapients that cannot become sapient. My rough answer to this is what I will call the halo effect. To explain this its helpful to I think about further cases in which there is some form of privation of rational agency. These cases, such as temporary insanity, involve something similar to the sandboxing we perform with children. Before and after such episodes the person can be counted as fully responsible for (and authoritative over) their commitments and their actions, but during them, we effectively downgrade the status of their agency, or at least hold it in abeyance. When we do this, we do not (or at least should not) cease to give these individuals the respect they otherwise deserve, but rather, our obligations to them are modified in certain specific ways. If anything, our obligations become greater, even if we give them less autonomy. Nor do we cease to identify the insane person with the sane person they were before, or become again later (there are cases in which this happens (e.g., brain death), but we won’t address them at the moment). It is a similar case with children, they do not at some point suddenly become a fully responsible and authoritative sapient being, and neither would they suddenly count as a different person upon doing so. There is an identity between the adult and the child they were, but one that is mediated by the sandboxing we carry out in child rearing.

What all of this shows is that there is not a crisp boundary between full sapience and mere sentience. The above cases present examples of practices that we have developed to police the fuzzy border between the two. The idea is that these are structural practices insofar as they play an indispensible causal role in maintaining our practice of respecting the autonomy of those we count as full fledged rational interlocutors. It is important to remember that a practice is a collective behavioural regularities, and thus that we are talking about maintaining and encouraging a certain tendency on our part. Now, I think that cases such as babies who will never develop anything resembling sapience, and previously sapient individuals who have irretrievably lost it, can be accounted for in this framework. I think that, loosely speaking, we simply can’t consistently maintain these structural practices (and thus the core feature of the institution of rationality as such) unless they extend to these cases. This is essentially what I mean by the halo effect: that the practices in which our respect of rational autonomy of sapients consists cast a halo of additional obligations to non-sapients in order to make them consistent. This halo is ultimately founded on the similarities that non-sapients share with sapients, and I think we’d find that the obligations are not homogeneous, but vary in relation to the level of similarity.

Now, there’s a lot of specific arguments to be had about what exactly ‘consistency’ is, and how this holds in specific cases, but I’m only trying to outline a strategy here, not solve the problem entirely (and I’ve just hit 6000 words). However, I think it’s helpful to say something about how this extends to the case of animals. As we’ve seen above, the scorekeeping abilities which form the basis of our capacity to engage in the game of giving and asking for reasons are based largely upon a practical grasp of one another’s behaviour, one that to a certain extent enables something like pre-linguistic communication (which Heidegger would still call ‘discourse’ or Rede). This same practical grasp extends to creatures that are similar than us, and it makes possible the ascription of derivative intentionality. Indeed, this can even be supplemented by a properly theoretical grasp of the similarities such creatures have with us (such as an account of the anatomical basis of pleasure and pain).

There is thus an automatic halo of cases in which our relations to non-sapients are directly based on the ways we relate to eachother, and it is gradated by the richness of such relations. The imperative to maintain the consistency of our practices then gives us certain obligations to these creatures on the basis of this richness. We have greater obligations to those creatures to which we can ascribe a vibrant ‘inner’ life than to those that we can simply ascribe rough states of pleasure and pain to. For instance, the fact that I can interpret my dog as having expectations about the future means that I must take that into account in the way I treat him in a way that I don’t necessarily need to with my goldfish. Nonetheless, there can be a general obligation to avoid needless cruelty in both cases. This probably also amounts to a hierarchy of value on the basis of the level of quasi-agency we ascribe to sapients, but I wouldn’t like to be tied down too much on how exactly that would work. I certainly don’t think it would be an absolute matter, whereby anything in the level above automatically trumps something in the level below.

In conclusion, I think I’ve demonstrated a possible strategy that enables us to address Jon’s worry in the context of something resembling a Brandomian normative framework. Importantly, I think I’ve given a far better spin to the Kantian idea Brandom referenced, whereby we are obligated not to act in certain ways towards non-sapients because of what it does to us, than the crude interpretation Jon proposed. It’s not that an act of needless cruelty does something specifically to me, such as make me ‘less sapient’, however that is to be understood, but rather how it reflects on our form of life as a whole. One might here maintain that if I could somehow causally isolate my act of cruelty, then it would suddenly become perfectly fine. I don’t think this works, because it flies in the face of the modified categorical imperative proposed above. We all have an obligation to follow the structural norms which maintain the institution of rationality, and we can’t isolate ourselves without placing the burden upon everyone else, which amounts to a lack of respect for their autonomy. Insofar as the causal dimension introduced by the modified categorical imperative has a certain consequentialist flavour, it is most definitely rule-consequentialism, rather than act-consequentialism.

As a final note, I think this strategy might strike some as reactionary, as a matter of simply trying to make Brandom’s account consistent with some prior conception of morality. I’d like to dispute this. On the one hand, I think much of what I’ve been trying to do is to make explicit norms that are already implicit in our practices. On the other, I think it’s quite possible that we can derive entirely novel obligations from the schema proposed above. I think the idea that we all have some transcendental obligation to the very institution of rationality itself could be particularly fruitful for developing a positive ethics. It definitely points in the kind of democratic direction that Habermas advocates. This will have to wait for another day, however.

Published by


Appropriate descriptors: (neo)rationalist, left-accelerationist, socratic wanderer, heretical Platonist, computational Kantian, minimalist-Hegelian, heterodox Foucauldian, dialectical insurgent, conceptual mercenary, philosopher of fortune.

12 thoughts on “Brandom and Ethics”

  1. Pete,

    This is a fantastic post. I’m reading his Heidegger essays next week, and I’m chuffed about the article on teleo-semantics.

    One thing I should have been clearer about is that I think ultimately Schopenhauer solved the myth of the given problem with his meditations on the body. We experience frustration, which is an experience of something being impossible, which makes it completely rational to understand the world in modal terms (“will”) without this being “projection” or “guilding and staining” just because the body we are experiencing is part of the world (and note that Brandom himself endorses some form of modified modal realism!). “The myth of the given” is just the result of Kantians not taking into account the experienced body as a material object. An analogous argument holds with respect pain. So for me Sellars/MacDowell does not lead to Hegel (though I agree with Brandom’s Hegelian model of concepts whole heartedly) but rather to a realist form of Schopenhauer. [Probably less defensibly, to the extent that it does lead to Hegel, I think it leads to the Spinozistic pantheism that MacDowell and Brandom distance themselves from.]

    This Summer I’m going to write a paper on Heidegger’s notion of world-richness (with MacIntyre’s conclusion that many animals are world rich) and try to show how this mediates between the false dichotomy Brandom sets up between rational choice theory and inferentialism, and also between mere categorization and inference.

    But look, the casual way Brandom equates parrots and rusting iron (both as mere categorizers!) without any citation of the voluminous empirical literature on animal cognition (see the anthology “Rational Animals?”) does stink. I’m reading his essay on cognitive science (at the end of Reason in Philosophy) in two days and I’ll post something at that point.

    I do think that these issues are separable from your ethical points though, which are also certainly worthy of a lot of thought. I’ve got to go to yoga now, but I’m excited about thinking them through later today.

  2. I’m afraid that you lose me very early on in this discussion. The passage where my intuitions come unglued is the one that begins with “There are a lot of ethical theorists who try to ground ethics in the intrinsic normative force of pleasure and pain…” and ends with “…we can find no good reason not to actively deceive others and ignore the wishes of the dead.” Perhaps if you could answer a few preliminary questions for me, I could follow you a little further down the Brandomish path. Here they are, then:

    1) Are you suggesting that the satisfaction/frustration of our wishes and desires is something that always _does_ have unconditional, underived value/disvalue? There are a legion of intuitively plausible counterexamples to this sort of view. One could begin with especially lurid examples from Freudian case studies, but there’s no need to stray this far from everyday common sense. Consider one time in your life when you were disappointed or harmed, not by the frustration of a desire but by the fulfillment of a desire that you later came to repudiate. How much temptation is there to say about such events “Oh well, it was at least good for me that my morbid/immature/pathological desire _was_ fulfilled?”

    Philosophers of an autonomian bent are often fond of distinguishing between “authentic” and “inauthentic” desires when faced with this sort of objection. Is this a direction in which you are tempted to move?

    2) Do you think that it’s _always_ wrong to “actively” deceive others? If so, I’d love do hear more about how you draw the active/passive distinction to get around all of the familiar apparent counterexamples to this claim (e.g. the Jews hiding in Kant’s cellar, etc).

    3) There are obviously all sorts of reasons to act according to the wishes of the dead that have to do with the effects that these actions will have on the behavior of living people. Do you seriously believe that the obligation goes beyond that, i.e. to possibly sacrifice the welfare (however this may be construed) of some x that presently exists for the sake of the wishes of some x that does not presently exist?

  3. OK, sorry, just one more question that arose for me just a bit further down. The move toward grounding normativity in utterly transcendental norms of rationality does seem to be the right way for the Brandomite to go if he wants to dodge some of the more spectacularly repulsive apparent implications of Brandom’s own claims about “sustaining the conversation” and whatnot. And this move does, it seems to me, remove the temptation to regard “intrinsic moral worth” (as you put it) as “some special metaphysical property possessed by a certain class of entities.”

    But then you seem to go right ahead and help yourself to the claim that such norms obligate us to respect the autonomy of “rational agents.” How can this happen? How do we even know that there _are_ such agents on the basis of purely transcendental considerations? To suppose that we do would be to sanction the inference that a being of a certain type exists merely on the basis of how it is defined. Ironically enough, Kant owes a good chunk of his fame to having pointed out the absurdity of such inferences (albeit only in theology.

    And even supposing that we can make this sort of a dubious Anselmian inference, how can the relevant transcendental norms of rationality be understood as having any specifically ethical content whatsoever, unless they also furnish us with at least some empirical (or, if you prefer, pragmatic) criteria for finding such agents somewhere within the sensory manifold?

  4. Okay, there’s a lot to respond to here, so I’m going to take it in turns.

    To take Jon first, amalgamating the issues from the comment here and the one at your blog, I think there are the following substantial points:-

    1. You think there if we accept Brandom’s idea about rationality primarily consisting in practical abilities to implicitly track inferential connections between claims and actions, then we are on a slippery slope which ultimately forces us to accept that animals commit inferences.

    2. You think Brandom’s equation of parrots and rusting iron in his main example is over simplistic, given the wealth of literature of animal cognition.

    3. You think Brandom’s approach needs to be supplemented with some account of how ordinary practical abilities are constitutive for content, and that some kind of schopenhaurian story is what we need.

    Now, I think you’re wrong on point 1. This is because you aren’t taking account of the distinction between ordinary practical abilities to engage in certain activities, which can involve very complex differential responses to situational factors and involve some kind of expectation, and a specific kind of ordinary practical ability, namely, to play the game of giving and asking for reasons.

    Although we might want to say that there can be general practical abilities for keeping track of the relations between different possible actions we can perform, this does not amount to the kind of implicit scorekeeping activity necessary to play the game of giving and asking for reasons. We need to be able to track the relations between assertions, and the relations between assertions and actions, in order to function as rational agents. It might be the case that the former kind of tracking plays a necessary role in the constitution of the latter, but this does not undermine the sharp distinction between the two. There thus isn’t the kind of slippery slope you think there is – to be counted as rational on Brandom’s account requires specifically discursive practical abilities that animals lack.

    On the second point, although we can agree that the example does simplify the situation, I don’t think this undermines the way Brandom uses it. The point of the example is to demonstrate the above point about what is required to count as a scorekeeper, and thus to have something like an understanding of the conceptual content of claims. The point was to show that the parrot doesn’t have the required discursive practical abilities to count as actually asserting something. The parrots differential responsive dispositions are indeed a lot more complex than the rusting piece of iron, the important point is that they are still not the right kind of dispositions to count as a discursive creature.

    All this being said, I still mostly agree with you on point 3. Brandom does need a better story about how the ordinary practical abilities we possess (and share with some animals) interlink with the specific practical abilities required for discursivity. I’m not sure the schopenhaurian root is necessarily the right one, but agree something else needs to be said here. I think the Heideggerian root is also promising, but I think the one issue we repeatedly run into is that of normativity. The Heideggerian story (as Brandom and others tell it) still requires our practical understanding of how to act in the world to be a normative matter (i.e., of how One should use a hammer). This makes it harder to connect it up with the practical abilities animals possess, though not necessarily impossible.

    I think overall, the issue boils down to what we think ‘understanding’ is. Brandom gives us a good account of what conceptual understanding is, but he only does it by appeal to a notion of practical understanding (shared to some extent with Heidegger) that itself isn’t adequately analysed. Getting clear about this seems to be priority number one.

    Moving on to Mark, I think most of your objections stem from misconstruing the obligations I see us as having to respect the wishes of others as unconditional. I’m really not a Kantian when it comes to the unconditional character of moral law. I probably wasn’t clear enough about this, but I think that these obligations are certainly defeasible.

    There might be times in which the right thing to do is not to respect someone’s wishes, be it for their own good, or because the good of others is at stake. Similarly, I wasn’t suggesting that we sacrifice the wellbeing of people now for the wishes of the dead, because the needs of the living trump them in most (though not necessarily all) cases. Indeed, there might also be cases in which the right thing to do is lie to someone, to tell them that their wishes have been satisfied, I just don’t think that the default is that it is always right to do so if it brings them pleasure. My point is simply that we can’t even make sense of there being such defeasible obligations on a purely sentience-utilitarian account.

    With regard to my supposed Anselmian move, I think you have the total wrong end of the stick. I didn’t specify precisely what the transcendental norms are, or what respecting the autonomy of rational agents consists in exactly, because neither does Brandom (at least not in full). However, the idea is that these transcendental norms give us obligations to the effect that IF we count something as having the STATUS of being a rational agent, then we must respect its autonomy. Rational agents aren’t a class of natural entities like cats or dogs. They are entities that we confer a certain normative status upon. The caveat to this is that I think we MUST count anyone as a rational agent anyone that we CAN so count. None of this says anything about the necessary of existence of beings that we can so count.

    Now, the criteria for who we can count as a rational agents are based primarily on who we can engage in the game of giving and asking for reasons, and thus based on our own practical ability to play the game. This doesn’t mean that we have no obligations to those who don’t speak the same language as us, because it is possible to extrapolate from our own ability to play the game criteria governing who would count as able to play the game in general, even if they can’t play it with us.

  5. Ok, your remarks on defeasability are helpful. I’m a little more optimistic than you are about the possibility of understanding the very notion of a defeasible obligation without a whole bunch of aprioristic machinery chugging away in the background, but I’ll save my (broadly Humean) monologue on that for another day.

    On the issue of the “STATUS” of being a rational agent, though, I’m still i the dark. I certainly don’t expect you or Brandom to provide me with a complete theory of the constitutive features of rationality. But the question I asked earlier just takes on a new form in light of your remarks here: viz., how can we _know_ when we’re engaged in the “game of giving and asking for reasons?” Such a “game” presumably requires at least one “player,” right? Or possibly even two, given Brandom’s apparent sympathy with certain aspects of the private language argument.

    If even this question sounds like it’s asking too much, I’ll settle for answer to either of the following. Is the determination that one is (or ever has been, or ever could be) involved in such a “game” an empirical matter to _any_ extent? You mention that in order to have this status a being has to be endowed with “discursive abilities” and I guess I sort of understand this – I do find it tricky to discourse at length about German opera with my chihuahua. It’s not that much easier with the kids in my Freshman classes, though. So for them, I guess, I should think in terms of potential rather than actual capacities. But how far down should this go? Why not to my toaster? If “discursive abilities” are what’s important, fine – I’m interested in the prior question of _how_ you think these such “abilities” get detected.

    As soon as one comes to articulate empirical criteria for (possible) rational agency, whether they’re formulated at the level of neurophysiology, or perhaps in some weird Chomskian idiom that appeals to the “poverty of stimulus,” how does one avoid the skeptic’s charge that our own evolutionary history has made our view of the matter far too parochial to be reliable detectors of beings with this sort of “status?” I had always thought that part of the motivation behind Kantian ethics (and autonomianism in general) was to give us some ethical guidance that we can follow even if this sort of empirical skepticism turns out to be well-grounded.

  6. Okay, I think you perhaps misunderstood my point about defeasibility. I wasn’t saying that it’s impossible to have defeasible obligations per se on a sentience-utilitarian account, simply that its impossible to have particular kinds of defeasible obligation. The point was that I don’t think you can give an adequate account of the _default_ obligations we have to other rational agents vis-a-vis their wishes, on the sentience based account. That these default obligations are defeasible is of no matter to this, it simply makes the claim that we do have such obligations more sensible.

    On the matter of how we actually do recognise others as rational agents, I will concede that there are some things that need to be worked out. As I noted above, we can recognise that speakers of foreign languages are able to play the game of giving and asking for reasons, even though we can’t actually play _with_ them. Your students are a less extreme example, wherein there are certain specific topics one can’t converse with them about, they simply don’t have inferential competence (and thus understanding) within certain parts of the linguistic field, and its fair to say that this will hold to some extent for everyone.

    As such, when we ascribe rational agency to individuals, it seems that we’re not ascribing them a grasp of any particular set of inferential norms, but rather a grasp of _some_ such norms. Underlying this grasp is a general set of capacities shared by anyone who does have such a grasp, namely, capacities to make moves within the game and to keep score on one’s own and other’s positions within it. This is just to say that they have a practical grasp of the transcendental norms discussed above. It should be noted that one can play the game with oneself, but only on the condition that one is capable of doing so with others.

    I think it’s important to say that someone who counts as a rational agent has a grasp of some inferential norms, and not just the underlying capacities, because this is the condition under which they could actually play the game with someone. It seems to me that when we ascribe the status of rational agency to speakers of a language we don’t understand, that we are committing ourselves to the possibility that we could in fact engage them in discourse if we developed a grasp of their particular inferential norms, or vice versa.

    The question is still how do we identify that someone does have the general capacities required for scorekeeping and a grasp of some set of inferential norms? The basic test is actually being able to engage them in discourse. Obviously, we know that it’s possible to mechanically fake conversation, so we want a fairly robust idea of engagement in discourse here. Although I wouldn’t want to be tied down, I’d say this is something like engaging in an open-ended conversation in which its possible to introduce new inferential norms. I think there’s possibly also a practical dimension to this, but I haven’t given the matter enough thought as of yet. Regardless, I don’t think this robustness is determined by any empirical criteria. Given this, it seems that when we ascribe rational agency to someone that we can’t actually engage in such a robust discourse with, that we are committing to the possibility of doing so if some additional conditions were met (namely, the acquisition of additional specific competencies).

    The question becomes, what criteria do we apply in these cases? This is where our biologically based and culturally modified abilities for understanding one another’s behaviour kicks in. These are indeed evolutionarily parochial. However, we are not restricted to them. We can extrapolate from empirical facts about our own capacities for engaging in discourse, such as our neurophysiology, and the structural regularities exhibited by our linguistic behaviour, so as to determine good reasons for counting something as something that we could potentially engage in robust discourse. The point I made above is that this doesn’t thereby amount to giving an empirical account of rational agency. What we get here is an empirical account of a variety of ways in which rational agency could be realised, but there could potentially be other ways in which it can be realised.

    At the end of the day, the acid test for whether some radically different form of intelligence counts as a rational agent is whether we can engage it in robust discourse, and the ability to do this will then enable us to expand our account of the empirical possibilities of realisation. I think that the intuitions behind this are the same underlying the turing test.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s