How to derive ethically appropriate recommendations for action? A methodology for applied ethics

We start from the assumption that one of the main tasks of applied ethics can be reconstructed as the application of basic moral norms and principles to solve concrete morally relevant issues in specific contexts. This means to develop concrete recommendations for action, i.e., to develop statements denoting specific realizable actions within certain contexts that prescribe what is (from an ethical perspective) deemed to be the right thing to do. For this purpose, three main conceptual steps are necessary. First, a basic moral principle or a general norm needs to be explicated. We refer to this as the A-level of ethical reasoning.²

A crucial problem is that basic moral norms and principles are, loosely speaking, too abstract to allow for establishing or identifying concrete recommendations for action in view of a specific situation. This is mainly due to their generality and absoluteness. Generality, loosely referring to Gewirth (1981), concerns the range of applicability of a prescriptive statement, i.e., to the question of which moral subjects should be guided by a moral norm or principle. Absoluteness, on the other hand, refers to the degree of a moral obligation. For instance, only very few obligations apply to any moral agent (regarding generality) and without any exception (regarding absoluteness). Some apply as permission towards some agents, some as prohibitions and some may be trumped by other obligations. Nevertheless, basic moral principles or general norms usually imply moral obligations of high generality as well as a high degree of absoluteness.

Against this background, it becomes clear that further steps are necessary to arrive at recommendations for action. In a second step of ethical reasoning, the generality and absoluteness of a basic moral norm or principle must be reduced to specify its prescriptions with a view to the context at hand. This results in moral judgments on the B-level of ethical reasoning, which conceptually specify the moral obligations formulated by a moral norm or principle on the A-level in view of certain general conditions, i.e., the general circumstances in which a norm or principle is to be applied.³ Such circumstances may be, for instance, conditions of poverty, environmental degradation, resource scarcity in health care systems, or pandemic situations. With a view to this, a basic moral principle needs to be specified in at least five dimensions in order to develop moral judgments: against the background of the initial normative standard, it has to name an addressee and a subject of moral significance or bearer of rights. It, then, has to specify the nature of the prescription, that is, the degree of obligation, permission or prohibition. Finally, it needs to detail the conditions under which an action is prescribed. In short, this includes to answer the questions of (1) who has to consider (2) whose moral rights to (3) what degree of obligation, permission, or prohibition (4) under certain conditions against the background of (5) a certain normative standard (exemplified by the basic moral norm or principle on the A-level).

However, defining a specific and realizable action is beyond the scope of moral judgments. To arrive at specific recommendations for action, it is, in a third step, required to include relevant information on the contextual determinants of a morally relevant situation. We understand a recommendation for action to be a normative statement that contextualizes the content of a moral judgment in view of a specific temporal and spatial context and, in addition, recommends a specific action that is realizable within this context (Dietrich 2009). Hence, recommendations in addition to providing rather general answers regarding the five dimensions of moral judgments aim at substantiating these aspects with a view to concrete real-world situations and contexts under the general conditions which motivated developing the underlying moral judgments. Recommendations for action thus aim at dealing with specific real-world problems in specific contexts and at suggesting courses of action that correspond to the obligations expressed in moral judgments. They are developed with a view to at least six questions: (1) who has to consider (2) whose moral rights (3) in which way to (4) what degree of obligation, permission, or prohibition (5) in a specific context against the background of (6) a certain normative standard (exemplified by a moral judgment on the B-level). Hence, when speaking of recommendations for action—in contrast to moral judgments—the additional dimension (3) is implied. Furthermore, as dimension (5) makes clear, it is not certain general conditions that are relevant for developing recommendations for action (as specified in dimension (4) of moral judgments), but rather specific contexts in which they are applied. Regarding the question of realizability, it is furthermore important to know, for instance, who actually can exercise what kind of action, what different measures might be available or what their effects and outcomes are. This, ultimately, is a matter of social organization, professional roles and responsibilities, causal effects, availability of resources and other, even more complicated issues. This third step results in what we call the C-level of ethical reasoning and requires extensive empirical knowledge on the actual context and needs to review what ought to be done in the light of what is.

This model exposes a problem that arises in the transition between different levels of ethical reasoning. Consider the different attempts to deal with the risks arising in the COVID-19 pandemic for residents of nursing homes. Furthermore, consider the basic moral principle that a person’s wellbeing is worthy of appropriate consideration: (a) represents the A-level of ethical reasoning. For the purpose of our argument, we will not define “wellbeing” further than necessary. For instance, we will leave open the question of whether wellbeing is best explained by hedonist, desire-satisfaction, or objective list theories (Parfit 1987). We assume, however, that “wellbeing” refers to the conditions for developing the affective, cognitive, and emotional capacities necessary for living a good life. We further assume that these—at least to a certain extent—include physical and mental health.

Obviously, many different moral judgments may count as a specification of a basic moral principle like (a) with a view to certain general conditions. One could, for instance, reasonably argue that (a) can be specified in ways that result in the following moral judgments: (b1-6) represent the B-level of ethical reasoning, i.e., they specify (a) in the context of a global (e.g., COVID-19) pandemic and with a view to the situation of residents of nursing homes. According to Paulo (2012), the relation of moral principles and specified instances of these principles is a “purely formal” one. The notion of “purely formal” denotes that it is to be understood as a relation between a class and its instances. While moral norms and principles, by defining general obligations to be realized as far as possible, constitute the defining properties of such classes, moral judgments constitute instances of these general obligations that satisfy their requirements given certain general conditions. However, as the examples of (b1-6) show, the process of specification may give rise to different moral judgments, with also differing degrees of abstractness or concreteness. However, not only are complementing judgments conceivable (as (b1-6) could be), but also contradictory ones. For instance, the claim that nurses are obliged to promote residents’ freedom of physical pain (b3) could, under certain circumstances, contradict the claim that nurses are obliged to promote residents’ mental health (b4), for the former could imply that nursing staff should reduce direct contact with residents to minimize the risk of infection whereas the latter may imply that contact should be increased to avoid emotional distress (however, this is not clear at the level of moral judgments).

However, it is not possible to rank such judgments according, e.g., to their moral value. Rather, any of the judgments (b1-6) must be prima facie considered equally valid with a view to the underlying basic moral principle (a) (if the relation between (a) on the one hand and the judgments (b1-6) on the other is formally correct). This is what we call contingency₁, designating the relationship between a moral principle and related moral judgments as not being necessary but not being impossible (and therefore not giving decisive reason to comply with a certain judgment). In other words, the relation between a moral principle and a certain moral judgment is contingent insofar as a moral judgment satisfies the general obligations of a basic moral principle (i.e., it is an instance of a certain class) but is not the only moral judgment conceivable that satisfies these very obligations (i.e., the given class has more than one instance). Thus, additional reasons must be given with a view to the question of why a certain judgment is preferable over others.

On some occasions of contingency₁, it might be possible to give purely theoretical reasons for one judgment being preferable, that is reasons for treating certain judgments as not being an instance of a given class. On other occasions, it might be the case that some moral judgments can be ruled out by competing moral principles, for example, by showing that they would violate other basic moral obligations.

(b1-6), however, are not recommendations for action. As regards (b3), for instance, it is not yet clear whether it implies that nursing staff should reduce direct contact with residents to minimize the risk of infection. As regards (b4), it remains open whether it implies that contact should be increased to avoid emotional distress. To arrive at recommendations for action it is necessary to further concretize moral judgments like (b1-6). For instance, it is possible to comply with (b3) in the three different ways mentioned in the introduction above by reference to differing contextualization, e.g.:⁴(c1-3) certainly fall within the scope of concrete recommendations for action. However, at first sight, it is not possible to rank these recommendations according, e.g., to their moral value. Any of the recommendations (c1-3) must be prima facie considered equally valid with a view to the underlying moral judgment (b3) (and hence with a view to the underlying moral principle (a)). This is what we call contingency₂, designating the relationship between a moral judgment and the related recommendations for action as not being necessary but also not being impossible (and therefore not giving decisive reason to take a certain measure). Instances of contingency₂ are of crucial importance in the realm of applied ethical considerations.

Deriving recommendations for action from moral judgments is not possible by reference to purely formal considerations as in the case of inferring moral judgments from basic moral principles. Rather, the relation between moral judgments and recommendations for action is of a more material kind, i.e., certain ways of referring to empirical evidence are necessary to develop concrete recommendations. Contingency₂ is primarily a result of the possibility to refer to different bridge principles in the process of inferring recommendations for action from moral judgments. A bridge principle is a non-trivial premise that—from the perspective of logical inference—together with at least one empirical premise and a moral judgment (Russ et al. 2006; Carnielli and Coniglio 2007; Mertz 2014) allows for concluding a specific obligation, permission or prohibition (e.g., a recommendation for action). Origins of the concept of bridge principles (or “bridge laws”, Nagel 1961) date back to the first half of the twentieth century and the debates around logical positivism. Aiming for a clear distinction of certain types of scientific sentences, logical positivists were faced with the question of how one type could be derived from the other. A way to explain this transition was to advocate for additional non-trivial premises encompassing features of both types of sentences (Carnap 1966). The idea was then prominently discussed by Hans Albert with regard to bridging the is-ought gap (Albert 1991). In the last decades, the concept has drawn increasing attention in the logic of ethical reasoning and bioethics (Schurz 1997; Russ 2002; Russ et al. 2006; Carnielli and Coniglio 2007; Paulo 2012; Mertz 2014; Romfeld 2019; Kuehlmeyer et al. 2022).

Since bridge principles merely describe a function within the reasoning process it is left to debate what exactly could serve as a bridge principle.⁵ For a recommendation like (c1), one could refer, e.g., to the Pareto principle, which (in the strong formulation) holds: Employed as bridge principle, (BP1) provides two empirically verifiable conditions for the claim that an action X that is for the good of some and does not harm anyone must be preferred over the alternative of not carrying out X. With regard to our case and existing empirical evidence it could be shown that, especially in confined spaces like nursing homes, asymptomatic infections are a driver of the pandemic (Ooi and Low 2020; Zhang et al. 2020). Strict epidemiological testing regimes including asymptomatic persons would allow to interrupt transmission at an earlier stage (Gostic et al. 2020). Earlier detection of asymptomatic carriers of the virus especially in high-risk groups would, thus, prevent at least some fatal outcomes. On the other hand, no evidence exists connecting tests for the coronavirus with any additional medical risk or harm for tested persons or for those who test, given suitable protective measures are put into place. With a view to moral judgment (b3) (and, hence, to moral principle (a)), the relevant empirical evidence as well as the Pareto principle, strict testing regimes could thus be recommended in a transparent way.

For (c2), it could be argued, e.g., by referring to a principle of comparable effectiveness. This principle states that (Raffée and Abel 1979): (BP2) states that those actions should be carried out by which a certain goal can be achieved better, i.e., more efficiently, than by alternative actions. If there is corresponding empirical evidence that the use of digital care technologies does not lead to a deterioration of care services and at the same time contributes to a reduction of contacts, this empirical evidence supports (c2) with a view to (b3) and (a). In this case, the use of digital care technologies would be a more efficient way since it is less associated with risks while resulting in identical outcomes. For example, it could be argued that tele-surveillance systems allow to observe and control routine care tasks without physical presence. Empirical evidence on some of these systems shows that the use of tele-surveillance systems is not associated with loss of quality of life in patients, can shorten hospital admissions significantly, and relieves care workers (Chopik 2016; Vincent et al. 2006).

As regards (c3), one could argue that concepts like healing presence, therapeutic use of self, intuitive sense, exploration of the spiritual perspective and patient-centeredness are at the core of nursing care. Conceptual analysis and empirical evidence show that almost all holistic concepts of nursing include the idea of an interactive process aiming at (re)establishing relationships, emphasizing positive aspects of the situation and nurturing self-care, thereby caring for the wellbeing of patients (Ramezani et al. 2014). Deriving (c3) would, thus, be to show that a certain task satisfies inherent goals of the functional role of nursing (Downing and Thigpen 1984) and must be pursued for this very reason. This is to employ what could be called a principle of telos stating that (Mertz 2014): According to (BP3) there is an obligation to act with a view to achieving a given telos. From statements about what certain practices actually are according to their telos as well as about ways to act accordingly, it is possible—in this view—to make evaluative statements about what someone ought to do as part of these very practices (Lawrence and Curlin 2011). In this case one might want to show that satisfying (c3) is an instance of caring for wellbeing which is connected to the inherent goal of nursing (and follows from (b3)) as well as (a)) and, hence, should be carried out.

It is important to understand that we do not advocate any of these arguments but rather use them for the purpose of demonstration. Furthermore, one may argue that there may be different recommendations for action which we failed to consider. However, the sole purpose here is to show how bridge principles and empirical evidence can be used to establish links between moral judgments (as instances of classes established by basic moral principles) and recommendations for action. We do not intend to develop substantive ethical arguments but aim at making evident that it is only possible to argue for or against certain recommendations for action, and, hence, to deal with contingency₂ if their relation to empirical evidence has been explicated.

Finally, under ideal conditions, one would perhaps want to conclude that all the recommendations (c1-3) should be implemented. In real-world situations, however, this is often impossible. For example, necessary resources might not be available. Therefore, ethical argumentation usually does not end at establishing recommendations for action. Rather, further empirical evidence or evaluation of the reliability of the available information could provide indications that a certain recommended course of action is preferable over others, for instance, because it comes at fewer risks. This part of ethical argumentation is important, but not part of the framework presented here. We do not deny the importance of this step but rather understand the proposed framework as a prerequisite since critically assessing strengths and weaknesses of arguments as well as adding further considerations can only be carried out after the preconditions of respective recommendations for action have been argued clearly and transparently.

In addition, the proposed framework is rather ideally shaped. For example, empirical support for the crucial premises will always be partial and only be gradually plausible rather than proven. This inductive nature of empirical knowledge seems to create a certain tension with normative reasoning as proposed, which usually proceeds deductively from general to more specific assumptions and seems to be less open towards questions of epistemic uncertainty or of a possible underdetermination of evidence for its empirical premises. However, the framework does not preclude considerations about the strength and weakness of evidential support but explicates its role for normative reasoning. Implications, for example, of inductive argumentation is, in our opinion, a separate topic that, now, can be referenced as having importance for the credibility of the empirical premises in a normative argument, which could be another step to improve rigor of argumentation in empirically informed ethics.

Our account of deriving ethically appropriate recommendations for action has several practical as well as theoretical implications for empirically informed ethics research. It could be claimed that contingencies may have serious implications for ethical recommendations as they may result in the impression of a certain arbitrariness of conclusions hampering their evaluation and adaption in practice. From a theoretical perspective based on our model of ethical reasoning, contingencies are an integral part of proceeding towards recommendations for actions that necessarily arise as the result of moral deliberation. However, we have also shown that overcoming these contingencies is far from being an act of arbitrariness but follows certain rules and necessities that can be derived from our model. While dealing with contingency₁ might be a rather general problem known to different branches of normative ethics, we have highlighted that dealing with contingency₂ is a specific task for applied normative ethics. For contextualizing moral judgments implies evaluating a certain normative requirement in light of a specific situation and, hence, requires, first, reference to relevant empirical knowledge as well as, second, introducing further substantial premises (i.e., certain bridge principles) to connect a moral judgment and the respective relevant empirical information.

This indicates the importance of bridge principles in empirically informed ethical reasoning, and at the same time raises difficult questions insofar as bridge principles in ethical reasoning have been largely neglected in theorizing about recommendations for action so far. We have defined bridge principles in accordance with existing literature as non-trivial premises that—from the perspective of logical inference—together with at least one empirical premise and a moral judgment allow for concluding specific obligations, permissions, or prohibitions. It must be noted, however, that there is no generally agreed definition of what constitutes bridge principles besides referring to its functional capacity of linking normative or evaluative with empirical premises in such a way that normative or evaluative conclusions become possible. This merely implies that bridge principles serve an important logical function. It does not, however, include any information or specification on how this function is served or what this implies for the quality and strength of a specific argument.

As our examples indicate, there seems to be a great variety of substantial additional premises that can be introduced as bridge principles (Kuehlmeyer et al. 2022) which, by their specific properties, substantially influence the recommendations one may be able to derive. Some of these principles may be uncontroversial, some may require more extensive scrutiny to show their soundness and acceptability (Kuehlmeyer et al. 2022). Some may, however, even carry additional normative weight and implications that might only be acceptable against the background of certain moral theories. A suitable example for this is the application of a principle of telos which allows to infer statements about what someone ought to do from what certain practices actually are according to their telos. This obviously implies understanding a telos in an Aristotelian fashion as something one ought to strive for. We, therefore, suggest that different bridge principles, depending on their properties may have different range, scope, and plausibility besides exercising their logical function as a connector. To use the metaphor of the bridge: it is by no means certain that all bridges are of the same size, quality, material, or durability. On the one hand, this shows that bridge principles decisively influence the strength and applicability of recommendations for actions while their theoretical structure and implications on the other hand are underexposed. The ambiguity and variety of bridge principles as well as their substantial contribution and importance to empirically informed ethical reasoning illustrates a gap at the intersection of is and ought that surely warrants more research regarding the structure and implications of different bridge principles and their use in ethical reasoning.

From a practical perspective, our considerations result in a shift of focus in terms of where the problem of contingencies and arbitrariness is situated. On the one hand, it seems to be quite usual that more than one recommendation for action can be derived in view of a specific morally relevant situation. On the other hand, however, it often seems to be the case that it is unclear how (by the use of what additional premises) these recommendations have been derived which, in turn, results in the impression of arbitrary recommendations for action. To our understanding, justifying and arguing for recommendations like (c1-3) involves not only explicating a basic moral principle like (a) and moral judgments like (b3), but also making explicit reference to that additional premise adopted to serve as a bridge principle. Given that transparency of scientific research and its reporting is a cornerstone of scientific debate, basic moral principles provide a point of reference for the further line of reasoning and help to establish a clear focus. Second, in specifying the formal relation between principles and moral judgments, it is crucial to be clear about the normative aspects in focus and possible alternatives. The general nature of basic moral norms and principles often allows for fundamentally different interpretations of values (like wellbeing or autonomy). The same applies to crucial conditions, addressees, and degrees of moral obligations. The use of bridge principles in justifying recommendations for action, however, poses the decisive link between empirical information and normative reasoning. It must be made explicit in research papers that combine empirical information and normative reasoning to allow recipients to understand the argumentation at hand (Kuehlmeyer et al. 2022). Finally, empirical evidence that is linked to the line of reasoning should be addressed in a precise way. The extent to which empirical data influences the reasoning via bridge principles should be critically discussed, and empirical hypotheses assumed to be true or testable in support of the argumentation should be stated. In this way, empirical research will be connected to other evidence and is opened for future empirical support or falsification.