Autonomy ought to be that of a universalizable principled maxim, not mine! EU AI Act. Ethical modest Prerequisites on Policy Decision Making.
Nikos Stasinopoulos (PhD)
Education Business Consultant, Researcher on Political Identities, Political Ethics, Justice, Political Psychosocial Research, Policies Foresight, under psychoanalytical and phenomenological oriented research methods
EU AI Act. Ethical modest Prerequisites on Policy Decision Making.
Autonomy ought to be that of a universalizable principled maxim, not mine!
Does ticking boxes equals informed consent?
A timely relevant, I think, elaboration and catch up with Onora O Neill?s Kantian version on policy making, delving into her technoethics.
Ι have a distinct, well versed, intuition that the crucial constituent in practical policy making is not knowledge ( which is in line with a dominant current according to which empiricists and cognitivists become allied bedfellows), but a kind of practical judgement that I will develop bellow.
When addressing ‘autonomy’, policymakers? across the political spectrum are usually focusing? more on an assumed -individual-Self that ordain, (self) legislates, disregarding the second part of this compound word, the? ethical, universalizable, principles. On the other side, if law is reduced to mere compliance to some guidelines is an inadequate audit mentality surely insufficient to achieve resilience for the citizens, and I think that if we remain on this mentality the postbreach cost in reputation and citizens’ confidence is far higher for a cyberresilient governance. Reducing the lawmaking principled morality to mere ‘right to privacy’ and right to choose’? remain in the? formal register of non-interference and not the civic resilience register of non domination of the people, of a non excludable ‘right to life in dignity’. Autonomy ought not be mine, but of a universalizable principle. Principled autonomy? based on reason and not on empirical preferences and inclinations,. Not individual autonomy with limited account of reason. Autonomy cannot in a resilient manner be 'Mine' in the empiricist's sense and constituents of the term.
In a nutshell, ethicswise and policywise, my severe reservation is that we cannot in a secure basis KNOW if there ever was a WORTHY ETHICAL act. Worth is not a matter of knowledge, but it demands cultivation and formation of the will. It is not a matter of former knowledge but, on our side a kind of CULTIVATED VOLITION to ENACT it. There are maxims of ethics and policy making prerequisites that are NOT COGNITEVELY, BUT VOLITIONALY incoherent when we try to ENACT them as UNIVERSAL guidelines. They can be coherent as a matter of knowledge but ethically unworthy in praxis.
In line with the current EU discussion on the upcoming Artificial Intelligence Act, I will?? bellow f address the issue of? autonomy and? ‘informed consent’. But, on? a first step, I will invite in a coherent and argumentative manner, policy makers’ to consider that their task is not to apply a concept or description to an act (which does not yet exist) –ethics of algorithms can accomplish it just as well-, but instead, to? exercise a practical? judgement and enact some norm or principle. Instead of applying a concept or knowledge, they have to enact ethical and universalizable practical judgements. Norms are not algorithms, and ethically choosing is not picking. How’s that?
?
The distinction between justice and beneficence is not sharply made within utilitarianism as a policy making principle. Utilitarian moral theory has then a rather paradoxical view of the value of human life. Living, conscious humans are (along with other sentient beings) necessary for the existence of everything utilitarians value. But it is not their being alive but the state of their consciousness that is of value. Hence, the best results may require certain lives to be lost--by whatever means--for the sake of the total happiness and absence of misery that can be produced.
On the other side, for the Kantian type of autonomy, avoiding being used as mere means and not as ends, does not imply to narcissistically? promote our inclinations? while being not accountable for our actions.
Kantians’ policy ethics does not claim to discover whether acts whose maxims they don't know fully are just. They may be reluctant to judge others' acts or policies that cannot be regarded as the maxim of any person or institution. They cannot rank acts in order of merit. Yet, the theory offers more precision than utilitarianism when data are scarce. One can usually tell whether one's act would use others as mere means, even when its impact on human happiness is thoroughly obscure.
Kantian policy ethics has a more restricted scope. Since it assesses actions by looking at the maxims of agents, it can only assess intentional acts. It sounds as if it is most at home in assessing individuals' acts; but it can be extended to assess acts of agencies that (like corporations and governments and student unions) have decisionmaking procedures. It can do nothing to assess patterns of action that reflect no intention or policy, hence it cannot assess the acts of groups lacking decision-making procedures, such as the student movement, the women's movement, or the consumer movement.
?
POLICY MAKING ETHICS: APPLY OR ENACT?
?I suspect, writing in applied ethics is mainly confident that the normative principles it invokes can be justified because it assumes that somebody else will provide the metaethical arguments needed if reasons for normative claims are to reach beyond the like-minded. However, even if we assume such a division of labour, and accept that applied ethics borrows rather than establishes ethical norms, a less obvious difficulty cuts deeper. This second difficulty is that it is not easy to see what it means to apply a practical principle, hence not easy to see how normative principles — even if they can be justified — can guide action.
I now turn to this second difficulty. The distinctive feature of normative work is that its direction of fit is unashamedly the converse of that of empirical or descriptive work that aims at cognitive truth claims. Its theories are neither explanatory nor interpretive. Rather than arguing that aspects of the world, among them representations, satisfy certain descriptions and principles, it argues that aspects of the world should be changed to fit certain descriptions and principles. That is what it is for work to be normative or prescriptive rather than empirical or descriptive.
?I do not think there is much disagreement on this point. But I think there is far less agreement about whatit is to apply normative principles, and how they are to guide action. The term ‘applied ethics’ suggests that normative principles are cognitively applied to particular cases, just as concepts or theories are applied to particular cases in the knowledge register of truth-oriented inquiry. However this is a misleading analogy. Writing in applied ethics depicts normative principles as applying to specific types of case or situation, rather than to particular cases. A focus on types of case seems both acceptable and unavoidable because the aim, after all, is not to take over the activities of practitioners in one or another domain of life by dealing with actual cases, but to suggest how certain sorts of activities might generally be well undertaken. So writing in applied ethics has to abstract from the details of actual cases, in favour of discussing schematically presented types of situation or case. Even discussions illustrated by reference to one or another well-documented example of a particular past case will see this case through the lens of specific descriptions — hence as a type of case. When ‘cases’ are discussed in applied ethics, they are seen as types of context or situation that may fall under a principle.
The more fundamental problem is that speaking of the ‘application’ of principles to ‘cases’, as is standard in some writing in applied ethics, suggests a misleading analogy with the application of empirical or theoretical principles to cases in epistemically naturalistic inquiries, where principles stand in need of reformulation, refinement, revision or rejection if they do not fit actual cases. By contrast in normative, including ethical, discussion agents who are committed both to specific principles and to specific accounts of cases or situations to which those principles apply, will have no reason to reformulate, refine, revise or reject a principle if a case or situation fails to fit the principle. The fact that way things are often flouts normative principles that we take seriously is a reason for seeking to change the world (in small part) so that it lives up to the principle, not for dropping or changing the principle.
In empirical, truth-oriented reasoning the relation of principle to particular case is correctly termed application, and principles must fit the cases to which they are applied, or be put in question. However in normative reasoning reference to cases merely indicates a type of context or situation in which a principle might be deployed (and even this point may have to be qualified). Empirical and theoretical principles are put in question by counterexamples that do not satisfy them; normative principles are not. Application of theoretical and empirical principles is essential to reasoning about the world, and can determine what should (provisionally) be believed. Application of normative principles is only a matter of identifying a context for use, and since principles are indeterminate does not determine what should be done. Normative, including ethical, principles are not deployed merely by identifying types of context for their use, but rather by enacting them. Normative reasoning aims to be action guiding. Its aim is to enact rather than to apply principles, and this cannot be done merely by specifying the types of context or situation for which a principle may be relevant.
Once we distinguish the application of principles from their enactment, additional questions come into focus. Given that principles, like the types of situations and cases to which they relevant, are indeterminate, they can have many possible enactments. Normative principles cannot determine just which of many differing enactments should be aimed for in a given situation. Indeed, given that multiple normative principles are relevant in most situations, it may be difficult to enact all of them, yet hard to assign priority among them. Do the indeterminacy and multiplicity of principles shackle their capacity to guide action?
On one view, any worry about indeterminacy may seem trivial. For some it is adequate that only if we imagine implausibly that principles are to provide algorithms for action, it might be said, will it seem problematic that they underdetermine their enactments. To enact a principle is simply to take some course of action that satisfies it; to meet an obligation is to act in some way that satisfies that obligation; to respect a prohibition is to act in some way that does not contravene that prohibition. Equally, to enact a multiplicity of principles is simply to take some course of action that satisfies them.
Yet, on reflection, this account of acting on normative principles says too little. It represents all ways of enacting a principle as on a par: but surely there are better and worse ways enacting any principle, or plurality of principles. Yet to say more about why some enactments are better than others we seem to need some additional point of reference. We need, it seems, not merely to justify certain principles, and to identify types of situation or context (‘cases’)for which they are relevant, but also some account of practical judgement that indicates how they should be enacted. Yet this has proved notoriously elusive.
These points might be taken — quite often are taken — as reasons to turn one’s back on any ethical position that centres on principles, and to look to some account of intuition or judgement of cases or situations, to carry the entire burden of practical, including ethical, decisions. I do not think this a promising route, although I shall say little about it on this occasion. I think there are broadly two ways to approach the problem that indeterminate principles do not show agents how to select among their possible enactments.
?One way would be to try to eliminate, or perhaps manage, the problem by formulating more specific principles. Any thought that problems that arise from the indeterminacy of principles can be eliminated or even managed by formulating more determinate principles seems to me implausible. Setting out more elaborated or specific principles does not eliminate indeterminacy.
On the other side, normativity requires indeterminacy because it requires relevance to situations that are still open and unresolved. There is no way of building so much into principles that indeterminacy is eliminated — and going in that direction by incorporating unending distinctions, restrictions and exceptions, thereby formulating more and more specific principles, is likely to have diminishing practical point. Any account of practical judgement needs to accept that normative principles are ineliminably indeterminate.
Practical principles, including ethical principles, are not deployed one by one. The types of cases that provide contexts for their use invariably fall under multiple normative principles, including multiple ethical principles. The fact that we are typically committed to numerous normative principles, including numerous ethical principles, that demand joint satisfaction can generate tension and conflict — even irresolvable conflict, so seemingly threatens to undermine the prospects for any principle-based approach to practical let alone ethical reasoning. I think that this thought is too hasty, and that commitment to a plurality of indeterminate principles does not automatically lead to irresolvable problems.
Agents who are committed to a plurality of principles often find that this creates no great difficulty. Nearly all of the time most agents manage to meet the demands of many well-entrenched ethical principles.
When we just merely apply empirical principles, contrary cases constitute prima facie counter examples and demand some refinement, reformulation, restriction or rejection of the principle applied. On the other hand, when we instead enact normative principles, it is unclear on what basis we view certain possible enactments as unacceptable or inadequate, as grounds for thinking that the (type of) case is ‘an exception’, or that the principle is defeated by some other principle. It looks as if viewing normative principles as defeasible? as a merely application, ends up endorsing at least a partial retreat from seeing them as genuinely normative.
?
Practical Judgement and Effective Policies
However, on another view, which I shall sketch, the defeasibility of normative principles is a strength, not a weakness, because it provides the basis for an account of practical judgement. By this I do not mean simply that it opens a way to escape from unwelcome aspects of rigid reliance on principles, as maintained by those who think principles irrelevant to ethical conduct. Those who doubt the importance of principles in practical reasoning do not see them as defeasible, but as dispensable; they cast doubt on the very need for normative principles, and throw the burden of justification onto responses to cases. Those who see principles as defeasible still take them to be important, and insist that even where justified principles cannot be enacted, they must be taken seriously. What they need to show is how agents who are committed to multiple normative principles are to deploy them. How is the move from a plurality of defeasible principles to one or another particular enactment of those principles be made? How does practical judgement work?
Reflective cognitive (which means that is knowledge based and not ethicopractical) Equilibrium is often seen as offering part of an answer to this question. It is a popular half-way house for those who find themselves uncomfortable both with the thought that there are indefeasible normative principles and with claims that principles are dispensable. By adopting a coherentist method that takes into account not only principles but ‘practical judgements’, it suggests that we can find a way both to adjust principles in the light of ‘cases’, and to shape enactments by principles.
?However, it seems to me that Reflective cognitive Equilibrium is really a way of seeking coherence among a set of principles of greater and lesser generality. ‘Practical judgements’ are not responses to particular cases (which can be pretty unconsidered). Rather they are normative judgements of types of cases — that is to say, they are themselves (more and less specific) practical principles. Reflective, cognitive Equilibrium in the end is a way of ‘equilibrating’, eg. seeking coherence? among a plurality of principles of greater and lesser generality.
However, an appeal to coherence among principles does not show enough about practical judgement. Often there are many ways of accommodating a plurality of principles coherently to one another, and some coherent accommodations may surely be better than others. Practical judgment should presumably offer some way of distinguishing better from worse enactments of principles. Reflective cognitive Equilibrium does not provide a way of doing this.? Nevertheless, Reflective cognitive Equilibrium offers a useful clue to the process of practical judgment. By drawing attention to the claims and the problems of living by a plurality of principles, it pinpoints the right starting point for an account of the enactment of practical principles. The task of practical judgment is to select among possible ways of jointly enacting the various principles to which an agent is committed. We never find ourselves confronted with the task of conforming to a single normative requirement.
The practical ambitions of ‘applied’ ethics are, I suggest, secured rather than undermined by the fact that we approach any situation with a plurality of defeasible normative requirements in mind. This is just as true of practical reasoning in which ethical issues are marginal, as it is of practical reasoning in which they are centre stage. Practical judgment is as much needed in bringing a building project in on time, on budget and to the specified design as it is in working out how to keep all promises made without unkindness. It is needed as much in working out how to respect standards of courtesy while conforming to the norms of contemporary business practice as it is in working out how care for patients while respecting their individual ‘autonomy’. These and countless other situations require practical judgement? about ways in which and the extent to which a range of ethical and other requirements can be jointly enacted.
This may seem surprising. I think this is because a large tranche of writing in applied ethics focuses on distinctive types of case where joint satisfaction of normative principles, and specifically of ethical principles, is contingently impossible, as in the literatures on moral dilemmas and dirty hands. This focus is readily taken as? if suggesting that commitment to a plurality of principles leads? supposedly only to conflict, even to tragedy. It may however be a mistake to focus too narrowly on these cases, although understandably tempting (especially for those eager to show the difficulty or even the incoherence of relying on principles). Of course, something will have to be said about the distinctive cases where joint satisfaction of principles is impossible and some of the normative principles to which an agent is committed — ethical or other — cannot be jointly enacted.
If we think of applied ethics merely as discussing principles and the types of situations in which they might be applied, we say too little about the practical task of working out how principles are to be enacted in those situations, and how conflicts and potential conflicts between them are best handled or averted. Nor do we learn much about this practical task merely by labelling it practical judgment. We could, however, learn a good deal about this task by considering the institutional and cultural means by which join enactment of a plurality of potentially conflicting principles might be eased, even institutionally routinised.
So practical ethics can go beyond the consideration of principles and the types of situations in which they could or should be applied by saying more about the institutional structures and cultural support needed if respect for significant ethical and other principles is to be adequately achieved in public, professional and private life. It could, for example, focus more on ways in which systems of accountability could be structured to support, rather than just underlining the loose placing and refusal of trust. Focus more on shaping institutional structures that can secure and allocate the obligations needed if human rights are to be taken seriously. It could say more about ways in which supposed rights to freedom of expression can — or cannot — be reconciled with intellectual property regimes. It could say more about the relevance — or inadequacy— of demands for transparency in improving communication and public policy. It could say more about better and worse ways of constructing regulatory regimes to secure and support compliance with ethical principles. These are rich pastures, and those of us who take normative reasoning seriously could explore them.
To do practical policymaking ethics we need not only to think hard about the justification of ethical principles, and the specification of types of or context for which they are relevant. We need to think less about with ‘application’, and rather address the implications of the fact that agents are always committed to a plurality of indeterminate and defeasible normative. Conflict and tension between enactments of principles may arise. In short, we need to move from discussions of principles relevant to specifiable situations and contexts, to discussions of policies and institutions that support processes of practical judgment by making it easier to achieve adequate enactments of those principles in more and more demanding situations. Commitment to a plurality of normative principles, I have suggested, does not undermine, but rather provides a basis for practical judgment. Practical judgment is exercised both in identifying ways in which normative principles can be jointly enacted — when they can — and otherwise in building and maintaining structures and policies that help to avert, limit or resolve tension and conflicts between enactments of normative principles, and in dealing with remainders where joint enactment is impossible or not achieved. Normative, including ethical, principles are not adequately deployed simply by identifying types of situation or case which fall under them, in which (not to which!) they could be applied. Their full deployment is a matter of constructing, supporting and working within structures and policies, institutions and cultures that buttress the feasibility of the joint enactment of the principles to which agents are committed.
For instance, we cannot consider Generative Artificial Intelligence as a decision making tool? that will be address? by us via? cognitive decision making inferential methods like Generative AI, just by duplicating? it and avoiding ethical decision.? The tool cannot? ethically decide to decide.
?
THE RITUAL OF INFORMED CONSENT
Having sketched above the inadequacy of automated, merely inferential cognitive decision making? reducing it to AI tools, we can now turn to the sensitive issue of? how these tools can nudge citizens to take predefined, by default,? actions. Also, are these Generative AI tools really ‘autonomous’? Can policy makers, experts or even citizens rely on these tools, using them to attract ‘informed consent’ on democratic decision making? Let’s delve into that.
Informed consent in? policies and dapplied ethics is commonly viewed as the key to respecting citizens’ autonomy. This claim is endlessly repeated but deeply obscure. There are many distinct conceptions of individual autonomy in circulation, and even more views of the value and importance of these various conceptions. In a survey of views of autonomy, Gerald Dworkin noted that it has been equated with: Liberty (positive or negative) ... dignity, integrity, individuality, independence, responsibility and self knowledge ... self assertion ... critical reflection ... freedom from obligation ... absence of external causation ... and knowledge of one’s own interests. Other writers have equated autonomy with “privacy, voluntariness, self mastery, choosing freely, choosing one’s own moral position and accepting responsibility for one’s choices”.The list could be extended in many ways, and the feasibility and the value of all conceptions of individual autonomy are hotly contested. It seems to me, however, that if informed consent is ethically important, this cannot be because it secures some form of individual autonomy, however conceived. Informed consent procedures protect choices that are timid, conventional, and lacking in individual autonomy (variously conceived) just asmuch as they protect choices that are self assertive, self knowing, critically reflective, and bursting with individual autonomy (variously conceived).
?Contemporary accounts of autonomy have lost touch with their Kantian origins, in which the links between autonomy and respect for persons are well argued; most reduce autonomy to some form of individual independence, and show little about its ethical importance. The ethical importance of informed consent in and beyond democratic practice is, I think, more elementary. It provides reasonable assurance that a citizen (subjected to AI tools) has not been deceived or coerced. I shall not rehearse the deeper theoretical reasons for thinking that we have obligations not to deceive or to coerce. I believe there are convincing reasons for thinking that we have such obligations, which provide good reasons not to impose action on citizens without their informed consent. In saying this I do not mean to suggest that informed consent is the only ethically important consideration. The libertarian tendency? technological ethics or even bioethcis sees informed consent as necessary and sufficient justification for action. For libertarians everything is morally permissible “between consenting adults”. Most other ethical positions do not view consent as sufficient justification. Even if there is informed consent, we may judge? for instance surgery without medical purpose, medical practice by the unqualified, or unnecessarily risky treatment unacceptable and may think it wrong to use human tissues as commodities, as inputs to industrial processes, or as items for display. Informed consent is one tip of the ethical iceberg: those who think otherwise overlook the rest of the iceberg.
How can consent show that there is neither deception nor coercion? What makes a ritual of informed consent effective? can a single signature—or a gesture of assent—imply consent to a range of distinct propositions? Proponents of specific and generic consent are at work up and down the land drafting regulations, codes of practice, and guidelines, consent forms, and information leaflets. Is it necessary to seek further consent whenever new research purposes are envisaged?
A reasonable starting point is to note that consent is a propositional attitude, given in the first instance not to another’s action, but to a proposition describing the action to be performed (other propositional attitudes include knowing, desiring, hoping, expecting, believing). Propositions may be more or less specific, and some limit has to be drawn to the amount of detail included. The inclusion of excessive or technical detail, for example, will eventually overtax even the most energetic, and undermine the possibility of informed consent.
On the other hand, consent that is too vague and general may also fail to legitimate action. It is commonly assumed that in consenting to a description of what is to be done citizens, lets say patients, also consent to other descriptions of the treatment or procedure that are—for example, entailed by or logically equivalent to the description to which consent is given. It is also commonly assumed that in consenting to a description of what is to be done the addressed citizen consents to the likely consequences of its being done.
Yet strictly speaking, consent (like other propositional attitudes) is not transitive. I may consent to A, and A may entail B, but if I am blind to the entailment I need not consent to B. Consent is said to be opaque because it does not shadow logical equivalence or other logical implications: when I consent to a proposition its logical implications need not be transparent to me. Transitivity fails for propositional attitudes. Consent and other propositional attitudes also do not shadow most causal connections. I may consent to C, and it may be well known that C causes D, but if I am ignorant of the causal link I need not consent to D. Again, transitivity fails for propositional attitudes. When I consent to a proposition describing an intended transaction, neither its logical implications nor the causal links between transactions falling under it and subsequent events need be transparent to me: a fortiori I may not consent to them. This I call the opacity of consent. Yet the logic of propositional attitudes suggests that we cannot simply assume that implied consent will spread from one proposition to another, or from one proposition to the expected consequences of that which it covers, making any further consent unnecessary.?
Our aim in seeking others’ consent should be not to deceive or coerce those on the other end of a transaction or relationship: these are underlying reasons for taking informed consent seriously. It follows that consent is not always improved by trying to ensure that is given to more, or more specific, propositions: more specific consent is not invariably better consent. Complex forms that request consent to numerous, highly specific propositions may be reassuring for administrators (they protect against litigation), and may have their place in recruiting research subjects: yet they will backfire if, for instance in medical ethics, patients or practitioners come to see requesting and giving consent as a matter of ticking boxes. Our aim should, I suggest, be to achieve genuine consent, and this may not always be best done by seeking specific consent to a great many propositons.? In the above medical ethics example, patients, research subjects, and tissue donors give genuine consent only if they are neither coerced nor deceived, and can judge that they are not coerced or deceived; yet they must not be overwhelmed with information. This balance can perhaps be achieved by giving them a limited amount of accurate and relevant information and providing user friendly ways for them to extend this amount (thereby checking that they are not deceived) as well as easy ways of rescinding consent once given (thereby checking that they are not coerced). Genuine consent is apparent where patients can control the amount of information they receive, and what they allow to be done.
?Genuine consent is not a matter of overwhelming citizens with information, arrays of boxes to tick or propositions for signature. Accurate information of varying degrees of specificity can be provided by offering fact sheets, explanatory leaflets, discussion, and (with care) by counselling—and time to absorb further information. If additional accurate information is reliably available as demanded, patients will not be deceived: even a citizen who decides on the basis of limited information has judged that the information was enough to reach a decision, and is not deceived. Nor is it difficult to give citizens greater control over what happens by making sure that their consent is rescindable, and that they know it is rescindable. Patients who know they have access to extendable information and that they have given rescindable consent have in effect a veto over what is done when it is technically possible) they remain free to rescind their initial choice. Where these standards are met, there are reasonable assurances that nobody is coerced or deceived.
ARE NORMS, ALGORITHMS?
On the surface norms seem defective guides to action. Norms, taken in the relaxed sense I have suggested, are abstract entities with propositional structure and content to which action is to conform. A norm formulates some standard or requirement, some recommendation or permission; action is then supposed to be guided or shaped by that standard, requirement, recommendation or permission. However, while norms are indeterminate, the acts done in living up to them—or in failing to live up to them—have to be determinate in all respects.
How, then, can norms guide action? Isn’t the temptation to think of practical matters, including morality, as a matter of being guided by certain norms an illusion, because norms are never enough to shape action? There is some temptation to think that this issue could be avoided if only norms were algorithms that provide wholly definite instructions for each context and can specify exhaustively what must be done in living up to them. This thought seems to me highly implausible—and particularly implausible for the case of ethically important norms: An algorithm is a finite procedure, written in a fixed symbolic vocabulary, governedby precise instructions, moving in discrete steps…whose execution requires no insight, cleverness, intuition, intelligence or perspicuity, and that sooner or later comes to an end.
?Strictly speaking, algorithms are therefore possible only within formal systems, where contexts and moves can be exhaustively specified., for instance, a system of arithmetic multiplication. But when we multiply in real life, we need more than algorithms:? There are no true algorithms for action. This may seem surprising. Does not Benthamite Utilitarianism ostensibly aspire to provide an algorithm for morality, supposedly allowing us to calculate which act is optimising? The formal structure of utilitarian reasoning may look algorithmic. The instructions for utilitarian calculation tell us to list all options; to reckon the probable outcomes of each; to calculate and sum the expected utilities of these outcomes for all parties; and finally to maximize.
In practice, as we all know, utilitarian calculation cannot even approximate the underlying algorithm: we can specify only selected options; we are often unsure about their probable outcomes; our calculation of expected utility for anyone (let alone everyone) is pretty gestural. Only the maximizing looks even close to algorithmic—however, on reflection we see that it too is not wholly tied down and could be done in various ways.
On second and more cheerful thoughts, it perhaps is not important if norms are not algorithmic. For they can at least formulate constraints on or advice for action, and perhaps all that matters is that we choose some act that meets the constraint or recommendation set by a norm that we are seeking to respect. Yet how are we meant to do this? A standard answer is that it is a matter of judgement. But this is not reassuring. Invoking judgement without explaining how it is to work seems to leave us no clearer about what we need to add to norms if they are to offer practical guidance. Yet perhaps practical reasoning can take us no further.
Or can we say something about how practical judgement is to go beyond norms and? select one rather than another enactment of a norm? the power of judgement is a special talent that cannot be? cognitively taught but only practiced.
One minimalist answer to this problem would be that practical judging is no more than a matter of lighting on some act that fits the norm under consideration. For example, if we are aiming to live up to a norm of promise-keeping we need only find some act that meets the terms of the promise previously made. Any such act will do, and there are no significant differences between the various acts by which a given promise might be kept. On such views, practical judging seems simply a matter of picking some act that satisfies the relevant norm, and it does not matter which of many differing available congruent acts is chosen. The original of such cases is illustrated by the predicament of Buridan’s ass, who could not find any reason for preferring one over another bundle of equidistant hay. Here, some writers want to say, choosing gives way to mere picking.
?There is no reason to think that the choice of one rather than another tub of margarine from a number that are equally accessible, equal in price, and indistinguishable in quality and appearance is more than a matter of picking. It is not a matter of choice because there is (ex hypothesi) no basis for choice and so no better reason for choosing one rather than another such tub.
But this example of mere picking is, I think, a limit case which we cannot take as a model for thinking about all cases in which a variety of discernibly different acts would satisfy the constraints set by some norm. Most ethical choosing is not a matter of mere picking. As Leibnitz thought, where things are absolutely different there can be no choice, because choice must have some reason or principle.? Even if we regard mere picking as a limit case of choice, it at least seems plausible that reasoned choice must refer to some norm. Moreover, there is often reason to think that the various ways of meeting the claims of a norm are not equivalent, and that some are better than others. Here it seems that we are not dealing with mere picking, but rather are deploying some form of reasoned ethical choice. If so, practical judgement must generally be more than mere picking, and there may be grounds for thinking that in a given situation with given norms, one judgement may be better or worse than others, and more generally that the judgements made by some persons might usually be better or worse than those made by others. This thought returns us to the problem of understanding how practical judgement works.
Practical Judgement and Multiple Norms
Is practical judgement an aspect of practical reasoning, or is it only a pompous term for acts of picking, by which we instantiate an indeterminate norm with any determinate act that fits the norm? If so, what is it that we admire in acts and persons that we think of as exhibiting good judgement? What makes one way of instantiating a norm an exercise of good judgement and another an exercise of poor judgement? Much of the literature on ethical judgement offers astonishingly little help here. Often this is because work that purports to be relevant to ethics and practical judgement is in fact about (one type of) merely theoretical rather than about practical uses of judgement, and in particular about the distinctive problems that arise for theoretical judgement when it is unclear which concepts or standards should be applied.
Kant divided theoretical judgement into determinant and reflective judgement, on the basis that: If the universal (the rule, principle, or law) is given, then judgment which subsumes the particular under it is determinate… If, however, only the particular is given and the universal has to be found for it, then the judgment is simply reflective. In both cases theoretical judgement presupposes that a particular aspect of the world is there to be judged. Where the judgement is determinant the task is to see whether a certain concept or description applies; where it is reflective the task is to see which of many possible concepts or descriptions is appropriate. None of this is relevant to practical judgement, where the task is not to apply a concept or description to an act (which does not yet exist) but to enact some norm or principle.
A large amount of writing on ethics (much of it Wittgensteinian in flavour) assumes, inadequately I think,? that practical judgement is form of reflective judgement, and so that it must address some version of the problem of relevant descriptions. For example, Peter Winch, John McDowell, David Wiggins, and at times Bernard Williams, depict judgement as the crux of the moral life, yet focus not on practical judgement but on judgement of the context or situation in which action is undertaken. They often describe ethical judgement as a matter of appreciating or appraising or attending to what is salient about situations and cases of ethical significance. A typical formulation is given by McDowell, when he characterizes judgement or deliberation as ‘a capacity to read the details of situations’ or a ‘capacity to read the details of situations in the light of a way of valuing actions’ or a ‘capacity to read predicaments correctly’. Yet the analogy between practical judgement and reading texts or appreciating situations is unconvincing. When we act we may as a preliminary matter have to decide how to view the situation in which we already find ourselves, and in which we seek to act: here reflective judgement may indeed be needed.
But even when reflective judging is completed, and we have determined how to view the situation, we will still need to decide what to do: and that is where practical judgement does its work. A focus on reflective judging will not reveal whether or how practical judging works.
If we think about action that conforms to a single norm there seems to be little that we can say about practical judgement. Any act that meets the relevant norm—that satisfies the standard or constraint it sets—seems to be as good as any other, and there is no room for distinguishing better from worse judgement. In this case we are indeed reduced to picking one of many possible acts that enact a practical principle. However, the thought of seeking to meet the constraints of a single norm is highly artificial. We constantly need to act in ways that meet multiple constraints and standards.
Practical judgement, I suggest, is not something different from acting on norms: rather it comes into play where and because numerous norms have to be simultaneously taken seriously and observed. Hence practical judgement can indeed be reasoned, because it is norm guided, and norms are apt for reasoning. If this account of judgement is correct, it follows that in the end there is always a point at which mere picking has to take place. When all steps have been taken to conform to the full range of norms—ethical and other—that an agent sees as relevant, there will still be a range of possible acts that fit all these requirements. Here reason giving has to stop, and here it is not needed. Picking comes into the picture when the differences between the possibilities from which the agent selects are indeed a matter of indifference. It is typically below the level of reason giving, and below the level at which any norms are relevant if I pick one rather than the other of two coins of equal value in giving change, or if I place my tub of margarine in one rather another equally suitable position in the shopping basket. Typically many of the aspects of action that fall below the level of attention or of intention are picked rather than chosen for reasons that are formulable in norms.
In offering this picture of practical judgement as reasoned, I have said nothing about the justificatory arguments that might be given in favour of some norms or against others. I have made no assumptions about the quality of arguments that we may be able to offer in favour of specific norms that are important in specific aspects of life, or about the possibility of offering arguments in favour of wholly general, unrestricted norms of ethical importance.? My question has been only whether if we had such arguments we would still find that norms were impotent to guide action, and I have argued that we would not provided we had reasons to think that a plurality of norms that are apt for reasoning make their demands, but that we could not expect those norms to guide us ‘all the way down’ to one rather than another particular act.
?
Some Conclusions
Practical judgment, I have argued, is a matter of finding acts (or policies) that meet the constraints of a plurality of norms that specify requirements and recommendations of a variety of types. However, the way that we deploy a plurality of norms is not best thought of as a matter of cognitive ‘balancing’ one against another. There is no cognitive metric for balancing or trading off different types of norms, and there are no reasons for thinking that high success in living up to one norm will generally compensate for failure to respect another.
Practical judgement is an aspect of practical reasoning because it aims to integrate rather than to prioritize or trade off a plurality of norms. This task can be done better or less well, and there are good reasons for thinking of some people as having good and others poorer judgement. But even those with the most intelligent and careful capacities for practical judgement will in the end have to pick among possible acts between which there is, as we say, nothing to choose. Provided picking is used only for this unavoidable task and not prematurely where respect for a range of norms has to be integrated, there will be no deficiency in practical judgement.
It follows that practical judgement will be at its most demanding when agents seek to respect multiple norms whose requirements are in tension, or even contingently? incompatible. Two strategies may be relevant here:
The first is a matter of forward planning and avoidance. We know a lot about the circumstances and actions that are likely to create conflicts between the requirements of different norms, and can try to avoid those situations. Those who make excessive or conflicting undertakings (bigamists, fraudsters) will not be able to honour all their commitments. Those who impose excessive or incompatible demands on others are likely to face them with impossible demands, which no exercise of practical judgement can integrate (the worst excesses of the target culture).
By contrast, foresight, care and good institutional structures can do a fair amount to avert such problems, by forestalling, reducing and averting contingent conflicts between principles, so easing the tasks of practical judgement. But individual foresight and social reform have their limits. Often there is no way of acting that satisfies all the norms than an agent would wish to respect, and nothing that could have been foreseen or done to avert the potential for conflict. The most extreme examples are ‘dirty hands’ problems, where institutions, practices and prior action make it hard or impossible for those who have to act to meet all the norms to which they are committed. Even when nothing is so deeply awry, we may often find that the background of institutions and practices, of habits and customs, of virtues and failings, of skills and incompetence, of capabilities and vulnerabilities which shape action also hinder attempts to live up to multiple principles. In making practical judgements it is pointless, indeed misleading, to assume these realities away.
It may be true that had institutions and practices been better, or had agents made better decisions in the past or been more competent, less conflict would have arisen. But in the world as it is, agents may be unable to avoid a degree of failure, including moral failure, because no amount of thoughtful practical judgement enables them to integrate and live up to all the norms to which they are committed in the situations that actually arise.
Where realities force hard choices it may simply be impossible for agents to meet all the norms that they seek to respect. The most that they can then do is to recognize the claims of unmet, contingently unmeetable requirements and recommendations. But the fact that a norm proves contingently unmeetable in some situations will not wholly cancel its claims. Unmeetable norms may leave ‘remainders’ that call for attitudinal responses such as expressions of regret or remorse, or for more active responses such as apologies, commitments to reform, to compensation, restitution and other forms of making good. The importance of active foresight and institution building, and of an active approach to dealing with remainders in the wake of practical conflict between norms and failure (moral and other) does not show that norms are redundant or useless. It shows that living up to them can be hard and demanding. Taking norms seriously is a matter of working towards practical judgement, expressed in action that seeks to enact requirements and standards and to make good where failure has not been avoided.
?