The Definition of Consequentialism: A Survey

Abstract

There are different meanings associated with consequentialism and teleology. This causes confusion, and sometimes results in discussions based on misunderstandings rather than on substantial disagreements. To clarify this, we created a survey on the definitions of ‘consequentialism’ and ‘teleology’, which we sent to specialists in consequentialism. We broke down the different meanings of consequentialism and teleology into four component parts: Outcome-Dependence, Value-Dependence, Maximization, and Agent-Neutrality. Combining these components in different ways we distinguished six definitions, all of which are represented in the philosophical literature. We asked the respondents which definition is best for consequentialism and for teleology. The most popular definition of consequentialism was the one which accepted value-dependence, but not maximization and agent-neutrality. We therefore recommend the use of this meaning to avoid misunderstandings. The results for teleology were more problematic, with several respondents claiming they never use the term, or indicating that it is confusing.

I. Introduction

While consequentialism is one of the most prevalent paradigms in ethical theory, the ways in which it is understood by different thinkers diverge significantly. This divergence is evident even among moral philosophers whose research focuses on consequentialism. The different meanings associated with the term ‘consequentialism’ cause confusion, and may be the cause of discussions which are not based on substantial disagreement but on different understandings of the terms being used.

All definitions of consequentialism accept the following condition:

Outcome-Dependence: whether something (e.g., actions, decisions, rules, character features, or any other thing or combination of things) is right or wrong (or to what degree it is right or wrong) depends completely on the outcomes that (may) occur.¹

The above is a necessary condition for any version of consequentialism. However, there are other conditions often associated with consequentialism on which there is no consensus. It is unclear whether these conditions are necessary for a moral theory to be consequentialist. They are:

Value-Dependence: whether something (e.g., actions, decisions, rules, character features, or any other thing or combination of things) is right or wrong (or to what degree it is right or wrong) depends completely on the goodness or badness of the outcomes that (may) occur.

Maximization: whether something (e.g., actions, decisions, rules, character features, or any other thing or combination of things) is right or wrong (or to what degree it is right or wrong) depends completely on whether certain things are (or may be) maximized.

Agent-Neutrality: whatever makes something (e.g., actions, decisions, rules, character features, or any other thing or combination of things) right or wrong (or right or wrong to a greater or lesser degree) must do so independently of the viewpoint of any particular agent.

The acceptance or rejection of the above conditions results in different definitions of consequentialism, and this has important consequences for categorizing a given view as consequentialist or non-consequentialist. The following are examples of some positions that may or may not be categorized as consequentialist depending on which of the above conditions your definition of consequentialism accepts.

Promoting things that are not valuable. According to some people, we should promote (either directly or indirectly) some things even if doing so does not promote what is good or reduce what is bad. For instance, consider the case of someone who thinks that the best situation is that in which aggregate welfare is maximized. Suppose that she also thinks that inequality should be reduced just because it is right to do so, even when doing so brings about scenarios that in her view are less valuable because they contain less aggregate welfare. This person would hold the view that something is right if and only if it promotes both equality and welfare. If our definition of consequentialism just accepts Outcome-Dependence, but does not include Value-Dependence, then this view counts as consequentialist. However, if our definition does include the requirement of Value-Dependence, then this view does not count as consequentialist.

In fact, this case can have a stronger version. This person could not just accept that we should promote equality and welfare, but hold the view that we should maximize both things. In that case her view would be consequentialist if our definition just included the requirements of both Outcome-Dependence and Maximization, but, again, not if it also demanded Value-Dependence.²

Satisficing. Some people hold the view that something is right if and only if it promotes (either directly or indirectly) what is good or reduces (either directly or indirectly) what is bad at least to some extent, although not necessarily beyond that. If our definition of consequentialism accepts Outcome-Dependence and Value-Dependence but rejects Maximization then this view counts as consequentialist; if, however, it requires Maximization, then it does not.

Agent-Relativity. Finally, some people hold the view that something is right from the point of view of a moral agent if and only if it promotes (either directly or indirectly) what is good or reduces (either directly or indirectly) what is bad from a viewpoint relative to that agent. They may defend this even though they can agree that this is not what would promote the best situation from an agent-neutral perspective. If our definition of consequentialism accepts Outcome-Dependence and Value-Dependence but rejects Agent-Neutrality then this view counts as consequentialist; if it requires Agent-Neutrality, then it does not.³

The acceptance/rejection of the above conditions results in confusion over whether a particular view should be categorized as consequentialist or not. Furthermore, it may also lead to unproductive discussions in which the parties to a dispute talk past each other. For example, many critiques of consequentialism emphasize its supposed demandingness, or its alleged inability to allow for preferential treatment for one's friends and family.⁴ Such criticism, however, is only relevant to consequentialist positions that endorse Maximization and Agent-Neutrality respectively, and so pose no threat to views that reject these conditions. To adequately assess the merits of any particular consequentialist position, it is necessary first to be clear on which of the above conditions that position accepts. Any strategy of criticizing consequentialism by assuming that all consequentialist positions must endorse some or all of the above conditions and arguing against those cannot succeed, as it is based on the faulty assumption that all versions of consequentialism must accept the same general set of conditions.⁵ Partly because of this confusion in the exact meaning of ‘consequentialism’, another term that is sometimes used instead of, or together with it, is ‘teleology’. However, the same problems we have seen in the case of defining consequentialism are present in the case of defining teleology.

We believe that the plurality of incompatible definitions of consequentialism and teleology causes confusion. This is regrettable and avoidable, because this plurality does not reflect a disagreement on any substantive philosophical problem. There is no one true definition of these terms. All the definitions presented here are reasonable and valid. The question of which of them we should accept is not a substantive question in normative ethics. In particular, whether we accept or reject Outcome-Dependence, Value-Dependence, Maximization, and Agent-Neutrality is something totally independent of whether we include the acceptance of those views as parts of the definition of ‘consequentialism’ or ‘teleology’. If the terms ‘consequentialism’ and ‘teleology’ did not exist at all, the debates on whether Outcome-Dependence, Value-Dependence, Maximization, and Agent-Neutrality are acceptable moral views could take place in the same way. Equally, these debates can take place in the same way regardless of whether we decide to use the terms ‘consequentialism’ and ‘teleology’ with one meaning or another. Our decisions about which meanings to give to the terms ‘consequentialism’ and ‘teleology’ don't raise moral questions that can make a difference in those debates. However, the fact that different people use those terms with different meanings adds confusion to those debates and significantly impedes their success. So, to facilitate substantive philosophical discussion it would be useful to have agreed upon definitions of both terms.

To investigate the question of how best to define consequentialism and teleology we ran a survey among specialists in this area. We did not rely only on what the most renowned experts in the field think. Rather, we intended to include a wide group of people who are likely to understand the implications of the different positions involved. The purpose of the survey was not to discover which normative views respondents considered to be correct, but to discover their opinions on the best way to understand the meaning of these terms. Furthermore, our goals were prescriptive, not merely descriptive. Our aim was not simply to learn how people actually understand these terms. Rather, we want to know which definitions of these terms are the best candidates to facilitate clear philosophical discussion and it is our opinion that the definitions favoured by a majority of the experts in the field are the best candidates to fulfil that role.

It is important to bear in mind that we are not claiming that there is some definition of consequentialism and teleology which, by combining the acceptance or rejection of Value-Dependence, Maximization, and Agent-Neutrality in a particular way, is better than others. Our working assumption is not that experts will be better situated to discover the right definitions of the terms in question, since, as indicated above, there are no such right definitions. There are competing incompatible definitions of these terms, all of which are valid ones. Because of this, we wanted to see if there were definitions for consequentialism and teleology that could garner more support than the others among those better situated to understand them. These could then be adopted as the standard definitions, thus avoiding confusion as different thinkers talk past each other. Finally, even if we fail to convince readers to agree on a particular definition, we think that there is value in pointing out the different ways in which thinkers use these terms – by making people more aware of the significant divergences in use of these terms, we may help avoid confusion even if no single definition is universally adopted.

II. Methodology

Questionnaire

We used a survey for quantitative and qualitative data collection using a Google form. The survey included two closed and two open-ended questions. With the latter we intended to gather further ideas we might not have initially considered that may be relevant in the definition of consequentialism and teleology.

The two closed questions were:

Which of these different descriptions of the meaning of ‘consequentialism’ do you consider more adequate?

Which of these different descriptions of the meaning of ‘teleology’ do you consider more adequate?

In both cases, we presented the same set of possible answers, which are listed below:

• 1. Whether something (an action, a decision, whether a certain rule is followed, the possession of some character feature …) is right or wrong depends completely on its consequences, agent-neutrally considered.⁶

• 2. Whether something (an action, a decision, whether a certain rule is followed, the possession of some character feature …) is right or wrong depends completely on the goodness or badness of its consequences, agent-neutrally considered.⁷

• 3. Whether something (an action, a decision, whether a certain rule is followed, the possession of some character feature …) is right or wrong depends completely on whether it maximizes the goodness and/or it minimizes the badness of its consequences, agent-neutrally considered.⁸

• 4. Whether something (an action, a decision, whether a certain rule is followed, the possession of some character feature …) is right or wrong depends completely on its consequences, whether agent-neutrally considered or from an agent-perspective.⁹

• 5. Whether something (an action, a decision, whether a certain rule is followed, the possession of some character feature …) is right or wrong depends completely on the goodness or badness of its consequences, whether agent-neutrally considered or from an agent-perspective.¹⁰

• 6. Whether something (an action, a decision, whether a certain rule is followed, the possession of some character feature …) is right or wrong depends completely on whether it maximizes the goodness or/and it minimizes the badness of its consequences, whether agent-neutrally considered or from an agent-perspective.¹¹

• 7. Other. Respondents choosing ‘Other’ were in each case invited in a different question to explain the meanings of the terms they considered more adequate.

We chose the six alternatives listed above because of the way they differ in their acceptance of the conditions required for a position to be consequentialist or teleological, as shown in Table 1.¹²

Table 1. Conditions for a view to be consequentialist/teleological included in each definition

Sample and recruitment process

We sent an invitation to fill in the questionnaire via email to 116 scholars on consequentialism. We selected them according to the following criterion: all the authors of articles in the ‘Varieties of Consequentialism’ section at the PhilPapers website with at least one citation. This was a way of discerning those who are most likely to understand the questions involved in the survey. We sent the chosen scholars a link to complete the survey in February 2021. To maximize the number of responses we got, we made the survey as short as we possibly could, and so we did not include questions about demography.

We obtained a total of 35 responses. Among them, there were 34 responses to the question concerning the definition of consequentialism and 33 responses to the one about the definition of teleology.

We considered the possibility of running another survey targeting a wider group of potential participants, including other scholars and philosophers who do not specialize in the specific topic of consequentialism, to compare the results with this one. However, we rejected this idea, given our prescriptive rather than descriptive aim. If we ran this survey targeting a wider group of non-specialists and the results conflicted, we would still consider more reliable the opinions of those who are more likely to understand the implications of the different definitions than the opinions of those who may not.

III. Results

The definition of consequentialism

Direct preferences

Among those who chose ‘other’, two of them presented options that appeared to accept Value-Dependence, and reject Maximization and Agent-Neutrality, according to which they could be considered close to definition 5 presented in the list (see Table 2).¹³ Another one accepted Value-Dependence and Maximization but did not address Agent-Neutrality, due to which they are closer to definitions 3 or 6.¹⁴ Finally, two other responses did not indicate any position on Value-Dependence, Maximization, or Agent-Neutrality, and provided other reasons to reject the definitions presented in the list, including a preference for defining consequentialism in terms of reasons for action, instead of what is right, and an objection that the definitions may not adequately accommodate forms of consequentialism other than act consequentialism.¹⁵

Table 2. Direct votes for preferred definitions of consequentialism

The direct votes for definition 5 were equivalent to 47.06% of the votes, which is a very high number. Moreover, if we count the two votes under the label ‘other’ that appeared to be close to definition 5 as equivalent to votes for 5, then that means that 18 respondents would have supported that option. This is equivalent to 52.94%, which is the majority of respondents.

Indirect preferences

By voting for a certain definition, respondents expressed approval or rejection of the different conditions which that definition, in comparison to the available alternatives, holds as necessary for a view to be consequentialist. Accordingly, we can extract from the responses presented above the stances with regard to the three conditions for a position to be consequentialist that are at stake (see Table 3).

Table 3. Support for each of the conditions for a view to be consequentialist expressed by direct votes for preferred definitions of consequentialism

In addition, we can also consider here the views held by those who voted ‘other’. These included two votes for Value-Dependence and against Maximization and Agent-Neutrality, as well as one vote for Value-Dependence and Maximization. Adding all these votes together, we get these results:

Total votes for and against Value-Dependence: 30 vs 2

Total votes for and against Maximization: 4 vs 28

Total votes for and against Agent-Neutrality: 12 vs 19

We can therefore conclude that:

There is substantial agreement that the condition of Value-Dependence is necessary to the definition of consequentialism: 93.75% of respondents agreed.

There is also significant agreement that that the condition of Maximization is not necessary: 87.5% of respondents agreed.

Most respondents also agreed that Agent-Neutrality is not a necessary condition either, even if the agreement here was not as substantial: 61.29% agreed.

These results coincide with the results of respondents’ direct preferences, suggesting that a definition like 5, which accepts Value-Dependence and rejects Maximization and Agent-Neutrality, should be preferred.

The definition of teleology

Direct preferences

Among those who chose ‘other’, three appeared to reject Value-Dependence and Maximization, and to therefore be close to definitions 1 and 4, despite being formulated differently from the ones in the survey (see Table 4).¹⁶ Another one rejected Value-Dependence and Maximization, and accepted Agent-Neutrality, in line with definition 4.¹⁷ Two others accepted Value-Dependence and rejected Maximization and Agent-Neutrality, thus presenting descriptions close to definition 5.¹⁸ Two of them just identified it with Value-Dependence, even rejecting Outcome-Dependence, thus holding a view clearly different from any among the six options indicated in the list.¹⁹

Table 4. Direct votes for preferred definitions of teleology

However, six other respondents stated that the term should not be used in this context. Two of them indicated that the term is not used by them or others. Three indicate that the term has other meanings different from the ones considered in this context, and one states that it is a confusing term that should not be used.²⁰

In this case, the direct votes that the most voted option got, which was again definition 5, were only equivalent to 24.24% of the votes. If we include the three votes under the label ‘other’ that appeared to be close to definition 5, then that means that 11 respondents would have supported that option, which is equivalent to 33.33%, exactly one third of respondents.

This figure is, however, closely followed by only one vote by the second most voted option, definition 4, both if we consider direct votes alone, where it receives 21.21% of all the votes, and if we include the three extra votes for it under the label ‘other’, in which case it would be equivalent to 30.3% of the votes.

Finally, we should bear in mind also that the respondents who stated that the term ‘teleology’ should not be used in this context were up to 6, which is 18.18%, not a negligible figure at all. In fact, it was the third most voted option.

Indirect preferences

In addition to the votes presented in Table 5, we can consider those under the tag ‘others’, which include three votes against Value-Dependence, five in favour of Value-Dependence, six against Maximization, and three against Agent-Neutrality. We thus get these results:

Total votes for and against Value-Dependence: 15 vs 12

Total votes for and against Maximization: 0 vs 25

Total votes for and against Agent-Neutrality: 4 vs 18

Table 5. Support for each of the conditions for a view to be teleological expressed by direct votes for preferred definitions of teleology

We can therefore conclude:

While most respondents seem to think of Value-Dependence as necessary in the definition of teleology, this is so by a small margin: 55.56% of respondents agreed.

It's clear that rejecting Maximization is necessary: 100% of respondents agreed.

Most respondents also agreed that Agent-Neutrality is not a necessary condition either: 81.82% agreed.

IV. Discussion

Consequentialism

According to the preferred option of the survey, consequentialism is the view that whether something is right or wrong depends completely on the goodness or badness of the outcomes that (may) occur, whether agent-neutrally considered or from an agent-perspective.

Support for this option was very clear. The second most preferred option, definition 2, was quite far behind the most preferred one, receiving only half as many votes. Furthermore, this definition was actually quite similar to the preferred one in that it accepted Value-Dependence and rejected Maximization (it only differed from the most favoured definition in its support for Agent-Neutrality). This did not really match what we found in the literature. One can easily find definitions that are compatible with this one in that they assume Value-Dependence and reject Maximization, but the explicit rejection of Agent-Neutrality in combination with this is not that common. In contrast, definition 3, which is very commonly used in the literature, got very little support, significantly less than we would have expected. There are different possible explanations of this.

First, it may be that many of those who accept Value-Dependence and reject Maximization also reject Agent-Neutrality, even if they do not consider it necessary to make this explicit in the definitions they use. This would explain why support for definition 5 is not that apparent in the literature. However, it would not explain why there is so much support for definition 3 (that accepts Maximization) in the literature.

Second, it may just be that the respondents to our survey were more careful to isolate the actually necessary features of consequentialism in their choice of definition. People simply using the term in their papers, without having been explicitly asked to give a precise definition of it, might be less careful to include only those features which, on reflection, they might consider to be necessary to consequentialism.

Third, the literature may just be reflecting the state of the art at different points in the past, that is, when the relevant papers or books were written. We suspect that if we had run this survey 10, 20, or 30 years ago the results would have been significantly different. This would mean that the discussions that have taken place in the last decades have influenced not just the normative views that theorists of consequentialism hold, but also the meaning they attribute to the concepts they use.

Fourth, according to another explanation, the familiarity of classical utilitarianism is putting a kind of backwards pressure on people's understanding of the term ‘consequentialism’. Since utilitarianism accepts Value-Dependence, Maximization, and Agent-Neutrality, and it is probably the most familiar form of consequentialism, it may be the case that people are inadvertently moving these aspects of utilitarianism ‘upstream’ into their understanding of consequentialism more generally. This might result in people inadvertently assuming that consequentialism must also accept these conditions, at least when they are talking loosely. This explanation is supported by Seidel's (2019) claim that the history of consequentialism can be understood as a process of ‘conceptual emancipation’ from utilitarianism, by which in successive ‘waves’ the various assumptions and substantive commitments of classical utilitarianism are modified or dropped, resulting in a diaspora of related but distinct positions. According to this account, the emancipation from the conditions of Agent-Neutrality and Maximization came later than the original emancipation from monistic hedonism and sum-ranking aggregationism. This would suggest that people have simply had more time to update their definitions of consequentialism to exclude hedonism and sum-ranking aggregationism as a necessary condition than they have had to update their definitions to reject the necessity of Maximization and Agent-Neutrality too.

Fifth, supporters of definition 3 may accept Seidel's story but prefer a different explanation of what is going on here. According to this, the desire to distinguish consequentialism from utilitarianism may have driven some philosophers to define it in ways that include as few features of utilitarianism as possible, thus rejecting both Maximization and Agent-Neutrality even while this was not necessary.

The fifth explanation would be at odds with the fourth one, and maybe, some would argue, with the second one too. But for the rest all these explanations are compatible with each other, so perhaps there are different factors concurring here which explain why the results of the survey do not match what one would expect by looking at the literature.

Finally, a different point that it is important to stress here is that the particular formulation of the definition of consequentialism that was chosen here is not really that important in itself. What matters is that the most chosen option is one that accepts Value-Dependence and rejects Maximization and Agent-Neutrality. This is why we recommend that whatever definition of consequentialism we prefer, it aligns with these expressed preferences. What defines consequentialism is that rightness, or reasons to act, ultimately depend on circumstances related in one way or another to the goodness and badness of outcomes.

Teleology

In the case of teleology the situation is more complex, for several reasons. Firstly, the number of direct votes for any of the six definitions offered for teleology was much lower than the number of votes for any of the definitions offered for consequentialism – 14 out of 33 respondents voted for ‘other’ instead of any of the 6 definitions offered. This alone suggests there is more confusion in the case of this concept. In addition, the most voted option for teleology received half as many votes as the most voted option for consequentialism. In fact, the difference between the first and the second most voted option was very close, just one vote. This was because it is rather controversial whether or not Value-Dependence should be a condition for a position to be teleological. Moreover, some responses even question whether Outcome-Dependence should be a necessary condition. Furthermore, a non-negligible number of respondents stated that the use of this concept is confusing, at least in this context.

Finally, there is another serious problem in that the preferred meaning for ‘teleology’ coincides with the preferred meaning for ‘consequentialism’. If this is so, then there is no point in having two different terms to name the same thing.

V. Objections and limitations of the study

Objections to our use of a survey

There are two main ways in which one might object to the approach we adopt in this article. The first is to question our assumption that, when it comes to facilitating clear philosophical discussion on whether we should accept Outcome-Dependence, Value-Dependence, Agent-Neutrality, and Maximization as parts of our definitions of consequentialism and teleology, the words we use to name different combinations of those views are not relevant, and what really matters is simply that we all use the same ones. It might be argued, if we reject this, that some definitions of ‘consequentialism’ and ‘teleology’ are better able to facilitate clear discussion than others, and the best way to discover which definition is best is not by the empirical method we have chosen (our survey) but by conceptual analysis. We nevertheless think that a survey is a better way of settling this question, for the following reasons.

First, we of course accept that conceptual analysis offers some important constraints on what counts as a good definition of a concept. Our use of a survey was intended to help us decide between competing, incompatible, yet equally viable and useful definitions. This is not the same as saying that just any definition of consequentialism would be reasonable as long as most people would agree to it. For example, a definition of consequentialism that rejected outcome-dependence would clearly be unacceptable, as would a definition which included utterly arbitrary conditions (e.g. that only consequences for people with blue eyes matter). Some degree of conceptual analysis then is necessary to formulate a serviceable definition. It is not sufficient, however, to determine the final definition that we should adopt. Conceptual analysis has been unable to narrow down the six definitions of consequentialism identified in the literature and rule out five of them as unacceptable, thus delivering us a single universally acceptable definition. The fact that philosophers have been analysing the concept of consequentialism since Anscombe coined the term in 1958 gives us some grounds for pessimism about the prospects of this being eventually achieved. We believe this is because, as stated before, none of these definitions is ‘true’ or better than others.

Second, though our method is an empirical one, our empirical method has not entirely supplanted the method of conceptual analysis as the choice each participant in the survey made to endorse one definition or another was not empirical. That is, participants chose the definition that they thought was best, and presumably they made their decisions by considering their own analysis of the concepts of consequentialism and teleology (though they may have also considered other things, such as how they think others understand consequentialism and teleology). Our survey allowed for conceptual analysis but also provided a method of bridging the gap between a plurality of acceptable definitions and a single agreed upon definition.

Finally, even if it turns out that we are wrong, and that (i) there really is a single best definition of consequentialism, and (ii) that definition can be discovered by conceptual analysis, it seems to us that the consequences of our mistake are less bad than the consequences of the confusion and breakdowns in communication caused by the plurality of definitions currently in use. If we are wrong, then we are overlooking a genuine substantive issue about how best to define the words we use, and at worst perhaps we discourage others from engaging in the necessary work of conceptual analysis that would eventually resolve the issue. However, the consequences of philosophers continuing to use a plurality of incompatible definitions of the same term seem worse – a slowing down of substantive work on the validity of important moral principles which make a genuine practical difference to questions of how we ought to act. Furthermore, if it does turn out that conceptual analysis is really the best way to settle on a final definition of consequentialism, then we think that there is some value in empirical work which clearly indicates the level of support that the competing definitions currently have.

The second main way of objecting to our approach is to accept that something beyond conceptual analysis is required in order to settle on a universally accepted definition, but to reject our approach as too restrictive. One might reject the idea that the best way to reach that agreement consists in surveying only people who are most familiar with the different meanings of the terms. There are alternative methods that we could have adopted. The democratic approach would involve surveying as many people as possible without limiting the pool of respondents by requiring that they fully understand the different implications of each definition. The dictionary approach would involve analysing a large number of texts that use the terms ‘consequentialism’ and ‘teleology’, extracting the definitions from each, and then suggesting that we adopt the most commonly used definition. Finally, the stipulation approach would involve us simply stipulating a definition of consequentialism and encouraging others to adopt it.

As indicated above, we chose to ask the 116 people selected by our sampling method not because we think that only they should have the right to have a say here, but because we can expect them to adequately understand the implications of, and the differences between, definitions which accept different combinations of Outcome-Dependence, Value-Dependence, Maximization, and Agent-Neutrality. These differences are nuanced and often difficult to see unless one is familiar with everything they imply. People who have been working on this topic are much more likely to fully grasp such nuanced distinctions. Other people (even other philosophy scholars or students), not having considered the implications of each definition before, may much more easily fail to fully understand what each definition implies. If we expected everyone to fully understand all the different definitions, then we would not be in favour of surveying only the experts. We would instead favour a survey targeting as wide an audience as possible, in line with the democratic approach.

The dictionary approach would present its own difficulties. First, it is often not clear which definition a given thinker has in mind when they use the terms consequentialism and teleology, as they do not specify which of the component parts of the definition they accept. Extracting definitions from texts would involve a great deal of interpretation, and there is a risk that we would end up interpreting texts so as to favour our own preferred definitions of the terms. Second, the sheer number of texts that mention consequentialism makes the project of analysing the texts and extracting a definition unmanageable, even if we were to sample only a small percentage of the total.

Finally, the stipulation approach seems especially unlikely to yield consensus, since it offers no reason whatsoever to prefer one definition over the others.

Problems with the options included in the survey

We have seen that there are reasons why some people may disagree with the preferred definition of consequentialism and teleology (definition 5) that are unrelated to whether or not they accept Value-Dependence and reject Maximization and Agent-Neutrality. One of the respondents indicated that we should define consequentialism and teleology in terms of reasons rather than rightness. Perhaps a good definition should cover both reasons and rightness, which the definitions we used in the survey did not, although it could be argued that one way of understanding the idea that something is right is to say that we have most reasons in favour of it.

Two respondents argued that the definitions we provided were committed to direct consequentialism or act consequentialism. In fact, we took this problem seriously when we designed the survey, and tried to avoid making this commitment in the definitions we gave. This is why, in all of the definitions, rightness and wrongness were predicated not just of actions but rather of ‘something (an action, a decision, whether a certain rule is followed, the possession of some character feature …)’. Still, this may have been confusing in other ways. We did not intend this list to include all the possible things of which consequentialists can predicate rightness or wrongness. Nor did we mean that consequentialists should choose one item alone from the list as the suitable kind of thing to be right or wrong – which would exclude global consequentialism. However, we understand that the list may have driven readers to understand that we meant that the list was exhaustive and disjunctive, even if that was not our intention. In addition, our list would have been more inclusive if it had featured ‘rules’ and ‘character features’, instead of items such as ‘whether a certain rule is followed’ or ‘the possession of some character feature’. According to some views, rightness and wrongness should be predicated of the internalization (Hooker 2000) or the teaching of rules (D. E. Miller 2021; T. D. Miller 2021), rather than of their following, and it may also be argued that it is exercising a certain character feature, rather than possessing it, that can be right or wrong (Driver 2001).

We could also have formulated the definitions as saying that ‘Whether something … is right or wrong depends completely on consequences’, instead of ‘on its consequences’. In this way we would not have excluded indirect consequentialist views. Some of these views claim that we should do what, iff certain special conditions apply, would bring about a good outcome, meaning that we should do that even if what we do does not actually end up bringing about such good outcome when such special conditions do not apply.²¹

Finally, it may also be pointed out that our formulation of the definitions, by being made in terms of whether something ‘is right or wrong’ (without any mention of degrees of rightness or wrongness), also excluded scalar consequentialism. We could have avoided this problem by making the definitions longer. However, we wanted to avoid overly complex definitions that might have discouraged people from participating in the survey.

We therefore understand that our efforts to provide comprehensive definitions may not have been successful. While we think that the definitions we provided are not committed to a certain type of consequentialism, we can see how they might have de-emphasized indirect and scalar consequentialist views. This being said, these objections do not seem to affect the results of the survey concerning its basic aim, which is to evaluate whether the conditions of Value-Dependence, Maximization, and Agent-Neutrality should be part of our definitions of consequentialism and teleology. We could perhaps have avoided issues with people disagreeing with the definitions provided in the survey for reasons independent of their agreement/disagreement with these three conditions by just asking directly about whether they agree with each of them. However, this might have increased the effort required to answer the questions, thus discouraging some respondents from doing so.

Another possible limitation of the study concerns the definitions we suggested for teleology. We could have included other options that, while not fit for consequentialism, may have been reasonable candidates for teleology. As Table 1 shows, there are two possible options we could have included in the survey but did not, namely those definitions that rejected Value-Dependence but accepted Maximization (there would be two versions of these depending on whether they accept or reject Agent-Neutrality). We did not include these possible definitions because while there may be reasons for rejecting definitions assuming Value-Dependence and for accepting definitions assuming Maximization, it seems that a form of consequentialism accepting Maximization while rejecting Value-Dependence would be odd. If we rejected the condition that consequentialism must be concerned with good rather than bad outcomes, it is not clear why we would accept the condition that something must be maximized in order for a view to be consequentialist. However, things may be less clear in the case of teleology, so perhaps we should have included those options as well. In addition, as we have seen, two people responded that teleology need not accept Outcome-Dependence. We did not include this possibility because we had not expected it as a possible answer, but given that some people did choose it, it seems to us now that we should have included it (though obviously not for consequentialism).

Finally, we also considered the possibility of including another question asking respondents about their favourite view. The purpose of this question would not have been to gather information about the issues we were investigating (i.e. whether the conditions of Value-Dependence, Maximization, and Agent-Neutrality ought to be part of the definition of consequentialism). Rather, the point of this question would be to mitigate the risk that the respondents would choose the definition that most closely matches their own preferred normative view, rather than choosing what they think is the most accurate definition of consequentialism. We decided against this because we believed that the risk of this happening was very low given that the respondents are experts in the field.

VI. Conclusion

The aim of this article was to discern whether the conditions of Value-Dependence, Maximization, and Agent-Neutrality are considered necessary to consequentialism and teleology, and so to settle on a way of defining those terms that would avoid confusions over how to categorize particular views, as well as avoiding more substantive confusions based on misunderstandings about how the terms are being used. In light of the results reached in the survey, we recommend using definitions of consequentialism that accept Value-Dependence and reject Maximization and Agent-Neutrality, regardless of whether they are formulated similarly to or differently from the definitions we provided in the survey. According to this, all that is required for a view to be consequentialist is that it satisfies Outcome- and Value-Dependence, regardless of whether it also satisfies Maximization and/or Agent-Neutrality or not. To use a broad definition, we can say that consequentialist views are those according to which rightness, or reasons to act, ultimately depend on circumstances related in one way or another to the goodness and badness of outcomes. We are aware that shifting to this definition may be uncomfortable for people who have been using other definitions for a long time. In fact, some of us running this survey have until now been supporters of a different type of definition including the condition of Agent-Neutrality. But we believe that the results of the survey should lead us to change the definitions we use, for the sake of improving philosophical practice in general and avoiding confusion. It might also be worthwhile for philosophers writing about consequentialism to explicitly state which conditions they are accepting or rejecting in their discussion.

As for teleology, the responses we got must lead us to recommend that this concept is not used in the context of normative views related to consequentialism. It seems to confuse rather than clarify things.²² Its preferred meaning just coincides with that of consequentialism, which makes its use redundant and potentially confusing. In addition, several respondents warned that the concept has many meanings and can therefore be misunderstood. Given this, if we continue to use the term we should be aware that we run the risk that people will misunderstand our intended meaning.²³