An Epistemic Argument in Support of Liberal Neutrality

My aim in the present paper is to develop a new kind of argument in support of the ideal of liberal neutrality. This argument combines some basic moral principles with a thesis about the relationship between the correct standards of justification for a belief/action and certain contextual factors. The idea is that the level of importance of what is at stake in a specific context of action determines how demanding the correct standards to justify an action based on a specific set of beliefs ought to be. In certain exceptional contexts –where the seriousness of harm in case of mistake and the level of an agent’s responsibility for the outcome of his action are specially high– a very small probability of making a mistake should be recognized as a good reason to avoid to act based on beliefs that we nonetheless affirm with a high degree of confidence and that actually justify our action in other contexts. The further steps of the argument consist in probing 1) that the fundamental state’s policies are such a case of exceptional context, 2) that perfectionist policies are the type of actions we should avoid, and 3) that policies that satisfy neutral standards of justification are not affected by the reasons which lead to reject perfectionist policies.


Article abstract
My aim in the present paper is to develop a new kind of argument in support of the ideal of liberal neutrality. This argument combines some basic moral principles with a thesis about the relationship between the correct standards of justification for a belief/action and certain contextual factors. The idea is that the level of importance of what is at stake in a specific context of action determines how demanding the correct standards to justify an action based on a specific set of beliefs ought to be. In certain exceptional contexts -where the seriousness of harm in case of mistake and the level of an agent's responsibility for the outcome of his action are specially high-a very small probability of making a mistake should be recognized as a good reason to avoid to act based on beliefs that we nonetheless affirm with a high degree of confidence and that actually justify our action in other contexts. The further steps of the argument consist in probing 1) that the fundamental state's policies are such a case of exceptional context, 2) that perfectionist policies are the type of actions we should avoid, and 3) that policies that satisfy neutral standards of justification are not affected by the reasons which lead to reject perfectionist policies.

AN EPISTEMIC ARGUMENT IN SUP-PORT OF LIBERAL NEUTRALITY
One of the defining features of an influential line within contemporary liberal political thought is the rejection of state perfectionism. Advocates of this line of thought posit that the justification of basic state policies should not depend on an appeal to religious, philosophical or moral conceptions of the good or comprehensive doctrines. 1 Rather, state policies that concern the main social and political institutions should be justified on the basis of neutral reasons or values. That is, they should prove acceptable to all affected parties regardless and independently of their deeper ideological disagreements. A commitment to find a neutral basis for justification would imply rejecting the basic tenets of perfectionism. And this insofar as it spells challenging a predominant notion in Western political philosophy 2 that sees the implementation and promotion of valuable conceptions of the good life as legitimate state action.
Perfectionism is attractive simply because it is based on notions that sound extremely plausible, at least at first sight. Every person is basically interested in leading a good life. If we conceive of the state as an institution whose aim is to help citizens to pursue their basic interests, it seems reasonable to presume that one of its crucial goals should be the promotion of valuable life styles and practices, and discourage those that prove trivial or harmful. Individuals develop conceptions of the good -in more or less articulate terms-that inform their projects. They often tap on wider religious, philosophical or moral doctrines to justify their beliefs. People normally hold that their beliefs about good life and the comprehensive doctrines they endorse are true or, in some sense, correct. Let us presume that we are firmly convinced that our conception of the good is correct and that such conviction is based on good reasons. We lead our lives as individuals or as members of various associations in civil society (say, a church) in accordance with such beliefs. Why should the situation change when we enter the political arena? An advocate of perfectionism might round up this argument stating that if we are actually justified in holding such beliefs and acting accordingly in the first case -when making decisions that concern our private life-we should also be justified to act on them in the second case -in political forum. To deny this seems to be a paradox or an inconsistency.
My aim in the following pages is, in the first place to defend the notion that a neutral state is morally superior to a perfectionist one. Secondly, to explain why liberal neutrality does not imply the presumed inconsistency mentioned above. In other words, to show that there is no inconsistency in believing firmly that our religious, philosophical or moral conception of the good is correct and, at the same time, admitting that it does not provide an adequate basis to justify fundamental state policies.

LET US BEGIN BY CONSIDERING THE FOLLOWING CASE:
Subject S1 arrives in a train station with the intention of boarding a train to city X There is a train at the platform due to leave in a few minutes, going in the direction of the city where the agent wishes to travel.
There are two train services with the same destination. The express train that takes 2 hours and another train that stops at several stations along the way and takes over 4 hours. S1 prefers to board the express train, yet she is aware that she will not undergo serious inconvenience if she boards the other train. She is planning to visit some friends but has not specified the exact time of arrival and does not believe her plans will be significantly disrupted if she is a couple of hours early or late. S1 reads the information board announcing that the train at the platform is the express train and overhears some people who confirm that information. She is persuaded it is the express train and decides to board it.
S2 arrives in the station at the same time as S1. She is also travelling to city X. In her case, however, it is extremely important to board the express train. One of her best friends will die if she does not reach her destination within 2:30 hours. Let us presume she must pay ransom to kidnappers, or that she is taking the only available medicine against a very rare poison with irreversible effects within 3 hours. S2 reads on the board that the train at the platform is the express service and she overhears the same persons as S1 stating that it is the express service. In view of these facts, S2 believes that it is actually the express train -she travels regularly along that line and normally such information would prove suff icient for her. Nevertheless, she soon realises that the consequences of a mistake would be profoundly serious and it is perfectly possible -albeit highly unlikely, in her opinion-that the information may be flawed. Consequently, she should try to get additional information that will support or refute her belief that the train she is about to board is, in fact, the express service. S2 thinks that if she fails to ask for additional information, having enough time to do so, her behaviour will be legitimately questioned both by herself and by others if things turn out wrong. Besides, she would be at least partly responsible for the tragic outcome of the situation. 3 Similar cases have recently been the subject of much debate in epistemology. Some authors refer to them as part of their arguments for contextualism. Defenders of this stance hold that standards of knowledge ascription to agents are context-sensitive. Stewart Cohen summarises it in these words, "the truth value of a sentence containing the knowledge predicate can vary depending on things like the purposes, intentions, presuppositions, etc., of the speakers who utter these sentences" 4 . Subjects S1 and S2 have exactly the same evidence; however, the ascription of the sentence "I know that p" (where p is equivalent to: "the train at the platform is the express train"), would be correct in the case of S1 and incorrect in the case of S2. Within the context where S1 is placed, in view of her goals and the limited importance of making a mistake, it would be perfectly sensible to judge that the information she has is sufficient to justify her holding that she knows that the train at the platform is the express service. Contrariwise, within S2's context, in view of the seriousness of the consequences entailed by a mistake, ascription standards go up. S2 should gather further information for the ascription to be justified. From this perspective, there would be no correct or incorrect standards per se, these would be established by the context, on a case by case basis. Jeremy Fantl and Matthew McGrath use these situations as part of their criticism of evidentialism and their defence of the stance they call "pragmatic encroachment" 5 . Evidentialism holds that the epistemic justification of a belief is determined solely by the quality of the evidence supporting it. As a consequence of this stance, if S1 and S2 have the same evidence to believe that p, either they are both equally justified or neither is. The case of the train would provide intuitive support to the notion that, contrary to the tenets of evidentialism, it is impossible to disregard the relevance of practical matters when deciding whether a subject's beliefs are justified or not. We saw that S1 and S2 have the same evidence, notwithstanding which, as a consequence of the different cost of making a mistake in each case, the former would be justified in stating "the train at the platform is the express train", the latter would not be justified. This would support 6 the idea that the justification of our beliefs and, in the long run, our knowledge, would always be determined by practical matters besides purely epistemic factors, that is, "truth-related factors" (such as evidence, reliability, etc.). 7 I will make a more philosophically modest interpretation of the implications of cases like the one described above than that offered by contextualists and advocates of pragmatic encroachment. In particular, the interpretation does not presume questioning evidentialism nor other widely common intuitions, both in orthodox analytical philosophy and common sense, that seem to conflict with these stances. It is certainly not my intention to criticise contextualism or defend evidentialism against pragmatic encroachment. A solid argument in favour of a political thesis as the one I expect to defend should be based on premises that may prove acceptable to common sense, as far as possible. 8 Let us go back to the case of the train. We can state that upon the basis of the evidence available S1 is justified to believe that p (that is, "the train at the platform is the express train") and to act in accordance with that belief, proceeding to board it with no further inquiries. As we have seen, S2 has the same evidence as S1 but, contrary to what occurs in the latter's context, a mistake would be very costly: a person's life is at stake. As we presume that S2 is able to obtain additional information that will strengthen her belief that p -or, alternately, have the opposite effect-acknowledging the high cost of a mistake associated to the context of her action should lead her to conclude that she is not justified to board the train based on the information available. The same situation may be posited taking only one subject in two different moments. Let us suppose that S1, in T, plans to visit some friends in city X. As in the original example, the subject prefers to board the express train. Upon reaching the station, she reads the information board and sees that the train standing at the platform is the express coach and hears some people confirming the information. She believes it is the express train and boards it. In T1, S1 finds herself in dramatic circumstances: somebody else's life depends on her catching the express train, she has the same evidence as in T, but she admits she is not justified to catch the train without further inquiries. This shows that it is perfectly possible -and, in fact, frequent-to be justified to believe and act on the basis of a belief in a given context C1, while one is not justified to act in a different context C2, which includes practical factors that were lacking in the first context. When the consequences of making a mistake are very serious, the mere possibility of being wrong dramatically increases the level of the standards of justification to be satisfied by the subject in order to act. In cases such as the example of the train, the mere possibility of a mistake causes justification standards to rise. The subject needs no concrete evidence showing that she may be wrong 9 . It suffices for her to be able to imagine a set of fortuitous and improbable facts, albeit not impossible, that might cause the opposite belief to be true. In other cases, some specific evidence would be needed showing that, although the scales are strongly tilted in support of her belief, the agent might be wrong. Gerald Gaus posits an interesting case with these features.
Sally the juror may firmly believe that Mort murdered Millie, but because there is some small but significant chance she might be wrong, and because the consequences of being wrong are so awful, Sally does not vote to convict Mort of murder. By no means should we take her vote as showing that she does not believe it -she may well avoid the company of Mort, whom she really does believe murdered Millie. 10 It is perfectly possible for a subject to be amply justified to believe that p (in this case, "Mort is guilty") but, the fact that there may be a slight, though significant, chance of being mistaken may well be sufficient not to justify her acting on the basis of that belief within a given context where her action may cause huge and irremediable harm to another subject. Notwithstanding that, this does not bar the subject from continuing to believe strongly that p and acting accordingly in other contexts where the risk of making a mistake is not equally grave. 11 Contrary to the case of the train, this type of situation requires that the agent have some sort of specific evidence that she might be wrong. Otherwise, and in view of the fact that making mistakes seems to be an inevitable part of any belief and action, no jury could ever convict a defendant who faces the possibility of serious punishment, which is clearly absurd. The point is that the standards of justification that rule in these scenarios are -and should befar more demanding than those deemed necessary to justify our beliefs and actions in situations where the consequences of making a mistake are not equally serious.
Other practical issues, combined with the cost of error, account for the gap between justifications of belief and action on the basis of that belief within various contexts. Justification standards are more stringent as a result of the greater importance of the issue at stake, but that is not the whole story. Different social functions imply special duties, which bring about various levels of responsibilities in case of damage to third parties. Let us pretend that somebody asks me if the lift of the building works properly. As I used the lift a few minutes earlier and it worked properly -it made no odd noise, for example, or unusual jerk-I reply that it does. I base my answer on the type of evidence that would normally justify this kind of reply. If the person who asked the question uses the lift a few minutes later and the machine fails causing the lift to collapse, nobody would blame me for the harm suffered by the subject or state that I acted incorrectly. However, if the person responsible for the maintenance of the lift had signed his monthly report certifying that the lift was working correctly on the sole basis of the kind of evidence I used, he would probably be sanctioned and might be held responsible for the damage caused by any potential accident 12 .
A further important practical issue, particularly when the consequences of a mistake are assessed from a moral perspective, is the imbalance in power and agency capacity of the subjects concerned. Let us suppose that agent A decides to do something whose effects have a direct impact on agent B. The result of an error made by agent A would cause serious damage to agent B. Nevertheless, in that scenario A would suffer no harm or, at least, nothing comparable to what would happen to B. B can do nothing to oppose or change A's decision. Briefly, B's situation is completely in A's hands, but A runs no significant risk as a result of her action. As we have seen, the higher the seriousness of the harm suffered by another subject if we make a mistake, the more stringent the required standards to justify our action. If the seriousness of the harm is compounded by an asymmetry as the one described above, the standards of justification will be even more stringent. There are imbalances both in the case of the train and in that of the jury. The fate of the person kidnapped (or poisoned), in the case of the train, and of the presumed criminal, in the case of the homicide trial, depends completely on the decision of an agent who suffers no direct harm if she makes a mistake. On the other hand, the subject has no chance to have any impact on a decision that may ruin her life. This asymmetry increases the degree of responsibility of the agent if her action is not adequately justified. Now, I have said that in contexts in which a combination of practical variables renders the cost of making a mistake extremely high, the fact that there is a significant chance of making a mistake is enough to have good reasons to abort our course of action -even when that chance is very small. In such cases, we should undertake courses of action satisfying more stringent standards of justification. The notion of "more stringent standards of justification for our actions" may be interpreted in several ways, depending on the context. In the train case, for example, agent S2 could satisfy more stringent standards of justification by simply gathering more information regarding the train waiting at the station. But there are other situations requiring another sort of response. Take the case of two scientists, A and B, who are working in order to produce technological applications of a same theory, which we can call T1. Now, although T1 is supported by strong evidence, insofar as it is a new, interesting scientific theory, the chance of mistake is always significant, however small it may seem to be. Was T1 mistaken, the projects of both scientists would fail. Yet, the costs of the mistake are different in each case. If A fails, no one will suffer any relevant harm. On the contrary, if B fails, many people will die or suffer important injuries. Therefore, in order to be morally justified to carry out his research, B must meet more stringent standards of justification for acting than A. Unlike the train case, in this context the fact that A conduces further research, that is, gathers more evidence supporting T1, will make no difference. Since T1 is an interesting, sophisticated scientific theory, the chance of its being mistaken is always substantive. And since definitive verification is impossible, nothing will modify this fact. What should B do to be morally justified to go ahead with his research? He may modify his project so as to drastically minimize the costs of making a mistake. B should render his experiments as harmless as that of A. Naturally, this may lead to another sort of problems: the technological device developed by him may become less efficient or worst in some other sense. Another alternative, which could be combined with the one just mentioned, would be to neutralize or diminish his own degree of responsibility in case of mistake. Such thing would be achieved, if, for example, all those potentially affected by the project agreed in a free, informed choice, to undertake the risks implied by the project. 13 Note that in this context the notion of "more stringent standards of justification" does not have an epistemic, but a practical nature: there is no gain in further researching; the aim is to reduce the chance of mistake and to obtain authorization to act on the part of the would-be victims.
The key intuition ruling this paper, to which I will refer in section II, is that the situation in the political sphere of a modern constitutional democracy is similar to the latter case. Modeling the institutions of the basic structure of society according to a comprehensive religious, philosophical or moral doctrine, constitutes in this vein a risky experiment that no one can be morally justified to carry out. Rather, a neutral state would be capable of satisfying the standards of justification required for action.
Let us summarise the main tenets developed above. 1. The fact of being justified to believe that p, under the assumption that it is a fallible belief, is not equivalent to being justified to act upon the basis that p in every context. Subject S may be justified in believing that p, based on very good reasons, that is, she may hold p with a high level of confidence but is not thereby justified to act on the basis of such belief within context Cx. 2. The gap between the justification of a belief and an action is a consequence of the weight of a set of context-dependent practical factors on the justification of actions. Central among these factors are: 3. The fact that the cost of commit a mistake when acting in Cx on the basis of the belief that p is dramatically higher than an error when acting on the same belief in C1, C2… Cn. 4. The fact that the subject acting in Cx, in this particular context, is in a position that enlarges her degree of accountability in the case of error. This may be due to b.1: the features of the social function she develops within that context (which may imply special obligations toward the subject affected by the action) or b.2: an asymmetry in power and agency between the subject deciding and the subject who would be harmed in case of error. When practical matters mentioned in 2 occur in a given context in a way that, say, makes the cost of error really very high, the mere possibility or a minimal but significant possibility of error constitutes a good reason for the agent to refrain from carrying out -within that context-the actions she would have performed on the basis of her belief that p in other contexts in which these practical matters had no place at all 14 . A justification of her action within that context would require compliance with more stringent and restrictive standards.

191
The meaning of the notion "more restrictive and stringent standards of justification" varies according to the characteristics of the concrete context of action in question. In some cases, satisfying such standards may simple require that people gather more information in support of their hypothesis. In other cases, this will not work. Instead, it will be necessary to drastically reduce the impact of the harm if a mistake takes place; to eliminate or diminish the agent's degree of responsibility; or both things.

1.1
As we shall see in this section, the exposition we have just offered will allow us to develop an argument in defence of liberal neutrality. state policies, in particular those that affect matters of basic justice and constitutional essentials -in Rawls' terminology 15 -are an extreme case of the type of scenario described above in the point 3.
Let us term C1 the context of action in which an agent or group of agents make decisions, on the basis of their beliefs about the good or the comprehensive doctrines they subscribe, about how to carry out their life-plans or the administration of the associations of civil society they belong to. Likewise, let us term C2 the action of the state when it enacts policies that significantly affect the basic structure of society. This includes the actions of state agents and the political participation of citizens. As was the case in the examples mentioned in the previous section, the passage from C1 to C2 involves a dramatic increase in the seriousness of the harm caused by error as well as in the moral responsibility of agents for such result if it should occur. In view of the change in the situation, justification of actions in C2 will require compliance with more stringent standards than those ruling C1.
Some of the factors that account for this asymmetry are: In the first place, the state is characterised for monopolising the legitimate use of coercion. Even though not all its actions are coercive 16 , coercion is essential to the institution itself. This is especially true about policies with the strongest impact on the basic structure of society. As Thomas Nagel correctly states, a state "is not just a cooperative enterprise for mutual advantage. The social rules determining its basic structure are coercively imposed: it is not a voluntary association" 17 . No other institution in civil society -churches, universities, political parties, professional associations, etc.-may exert an equivalent authority and power over its members.
Secondly, the state can offer a full array of primary goods that constitute the necessary conditions for people to enjoy the possibility of developing the plan of life they deem valuable. Such basic goods are, among others, a set of civil and political freedoms, opportunities for equal access to the various social functions and occupations, the provision of certain economic and material resources, etc. No other association in civil society is able to guarantee these goods. In fact, some of them -such as political and civil freedoms-are necessarily out of their reach. The state has control over the army, police, public education; it has the capacity to control the media, to enforce policies that impact on the global system of property, language, immigration, etc. Needless to say, no association in a democratic society can boast comparable social and cultural resources. No other association has a comparable impact over the life of citizens: it may create, as no other institution can, the adequate conditions for every one to have the chance to lead a good life and, obviously, it can also cause devastating harm.
The sometimes irremediable harm potential of coercion on the well being of individuals and on their capacity to pursue their life goals and projects imposes standards of justification in view of the potentially affected. Such standards are more stringent than those required from decision making processes within contexts of interaction where such possibility would be inadmissible. Similar considerations apply in connection with the decisive interference of the state in the availability of the primary goods mentioned above.
An equitable availability of those "all purpose means", in Rawls' words, is a necessary condition to develop any conception of the good that citizens may subscribe. In view of the fact that no association or restricted community may guarantee the availability of those goods which, on the other hand, are indispensable for such associations and communities to exist, we may gather that the distribution criteria of such goods should be more stringent than those required for the distribution of goods sought after by individuals within those associations and communities. After all, any individual in a democracy can leave associations and communities without losing access to any of the primary goods that allow her to make other choices: such as joining other associations, changing her beliefs and style of life, etc. Contrariwise, the loss of primary goods whose availability is guaran- teed by the state causes the subject to lose all freedom of action and leaves her with no options (with the exception of the costly alternative of emigrating, which is not always feasible). Thirdly, the coercive character of state activities -or at least of a very significant portion of them -and its control over citizens' access to primary goods also results in a dramatic increase of the level of responsibility in case of error. Let us presume, for example, that a Catholic priest thinks that the decision of church authorities to demand celibacy from priests is unfair and wrong from a religious perspective and that it causes him severe harm. Even if the priest were right and church authorities were making a mistake, the degree of responsibility of the hierarchy of the Catholic Church for the harm endured by the priest would be incomparably lower than the responsibility of the state when its policies cause significant damage to a citizen's wellbeing. As we have already pointed, neither the Catholic Church nor, obviously, any other association within a contemporary democracy has the right to coerce its members or restrain their capacity of access to the primary goods necessary to change the course of their lives and try to fulfil their religious vocation or, more generally, their ideal of a good life through other means. Since the priest is free to leave the Catholic Church and lead the life he pleases, if he accepts to remain within the Church and endure the alleged injustice, he does so voluntarily and, consequently, he can be said to self-impose the pain it entails. Therefore, the responsibility of the authorities of the Church for the harm caused is neutralised or, at least, significantly minimised. The situation is radically different when it is the state that commits an injustice that seriously damages the wellbeing of a citizen. In view of the coercive character of its power and its capacity to control access to primary goods, the subject will normally lack any chance to avoid the action of the state and the ensuing harm. The fate of the agent is in the hands of the state. Asymmetry in power entails a corresponding asymmetry in responsibility, which obviously falls upon the state. It is also accompanied by the need to comply with special justification criteria, far more demanding than those that can be reasonably required from people who lead or run other types of institution.
These remarks confirm the idea that the political sphere is a context in which, due to the very high cost of error, the scale of its potential damage to third parties and the high degree of responsibility imposed on agents by the ensuing results, the mere possibility of falling into error suffices to conclude that justification for action should comply with more stringent standards than those required in other spheres of interaction. There is no equivalent to the factors mentioned above when a subject makes decisions concerning his plan of life, or participates in the administration of associations in civil society. Consequently, justification standards deemed correct within those contexts cannot be expected to be appropriate in the political sphere too. An agent who fails to provide an adequate reply to the factors that create asymmetries between C1 and C2, by satisfying more stringent standards of justification than those that would justify her action in non-political contexts, fails to show an appropriate concern for the basic moral commitment to cause no harm to other subjects. At the same time, this means that she fails to comply with the demand to offer an adequate justification for her political proposals to affected parties (or, at least, to those who reject her proposals). This requirement of justification is one of the basic duties and essential conditions of political legitimacy of actions in a contemporary democratic state. The essential policies of a state should be open to justification based on good reasons whose weight and appropriateness may be acknowledged by all affected parties. In negative terms, policies that rule the basic structure of society should not be liable to reasonable rejection by any of the affected parties. If this is not the case, the state or the citizens who advocate those policies fail to honour their basic commitment to acknowledge the equality of all citizens. A crucial component of treating everybody with equal consideration and respect is to honour the commitment to offer a justification of the policies that shape the basic structure of society that will prove acceptable by all parties. 18

1.2
Having said that, a comparison between a perfectionist state and a liberal neutral state shows that the latter is the only one with a real chance to offer an adequate reply to the increase in stringency of the standards of justification for action that characterises the passage from C1 to C2. Contrary to the case of a perfectionist state, a liberal neutral state minimises a) the probability of error, b) the scale and seriousness of the damage and, c) the degree of responsibility of agents involved in this result in case it actually occurs.
Let as choose as a model of a neutral state, one whose basic policies are ruled by an interpretation of the principles of justice posit- ed by Rawls' theory 19 . Such interpretation is enacted by the various successive majorities along the democratic process, through the development of different political conceptions of justice. In Rawls' view, political conceptions of justice have three basic defining features. First, although they are moral conceptions, they have been worked out to be applied to a specific subject: the basic structure of a democratic society. Second, accepting such a conception does not imply any type of commitment to any comprehensive religious, philosophical or moral doctrine in particular. Third, such conceptions are formulated, as far as possible, only in terms of fundamental ideas belonging to the political culture shared by the community and are therefore familiar to all citizens. 20 Rawls introduces three additional features which account for the specifically liberal character of the political conceptions of justice: these conceptions a) enumerate rights and liberties of the kind familiar from a constitutional democratic regime; b) assign these rights and liberties a special priority, particularly with respect to the claims of the general good and perfectionism values; c) assure for all citizens the requisite primary goods to enable them to make intelligent and effective use of their freedoms. 21 A political conception of justice makes no general statements on the nature of justice. It merely stipulates principles aimed at regulating basic social institutions but does not expect such principles to be valid in other contexts (such as interpersonal relationships in private life). In the same vein, it takes no stance on valuable or non-valuable lifestyles, nor does it reply to questions related to the global sense of our existence. Consequently, the normative and evaluative content of a political conception of justice will always be lower than the content necessary to justify the policies of a perfectionist state. The latter should include, at least, some ideas on the notion of a good life. A relatively obvious consequence is that the probability of error of a political conception of justice is lower than that of a more encompassing doctrine. Increased content necessarily entails a proportional increase in error-probability 22 . Let us suppose subject 1 says "there will be snow in X tomorrow" and subject 2 states, "there will be snow in X tomorrow morning", and subject 3, "there will be snow in X tomorrow morning during one hour and ten minutes". Subject 2's sentence includes more informative content than subject 1's and subject 3's more than subject 2's statement. Needless to say, error-probability increases in the same way. Potentially, more things could refute subject 3's statement than would be the case with the statement of the other two agents. This also applies to the statement of subject 2 as compared to that of subject 1.
Anyone who believes -as perfectionist conceptions seem to take as a standard premise-that religious, philosophical or moral conceptions of the good have cognitive status, that they can be either true or false or, in some sense, correct or incorrect, should acknowledge a similar pattern. Increase in content -evaluative or normative in this case -should imply a corresponding increase in error probability.
The greater the span and articulateness of a conception of the good, the more it says when defining a good life, the higher the probability of containing errors. Such is the case too when comparing the model of a neutral liberal state just described to a perfectionist liberal state. Let us presume, for example, that the latter combines Rawls' two principles of justice with a set of moral ideals associated with liberalism such as autonomy, the value of experimentation and of a critical attitude in the face of community traditions. It also enacts certain policies whose aim is to promote these beliefs and life-styles and discourage those incompatible with them. When we compare the neutral state to the perfectionist liberal state, we find a widening of the corpus of beliefs at stake in the latter and a parallel widening in the range of actions that might be justified and carried out because of the now larger scope. The same simple reasons mentioned above, the probability of error and, in that case, of damage to affected parties, will be higher in the perfectionist liberal state than in the neutral state.
Given the fact that it is impossible to cancel the possibility of error, there is no way to guarantee that a neutral state will not commit errors, be they evaluative or connected to merely factual factors involved as, say, in the enactment of policies or institutional design. The level of interference in civil society within a neutral state will be significantly lower than in a perfectionist state. A neutral state is only committed to guarantee the availability of such primary goods as are necessary for all citizens to have real and equitable chances to carry out their projects. If a perfectionist state, no matter how moderate, expects to differ in actual practice from a neutral one, it will need to cause a wider and deeper effect on civil society. Success in achieving its goals -whatever the means used to do so: coercive or non-coercive -requires that as a result of its policies there will be an increase, as compared to the previous sit- uation, in the number of people who subscribe to and carry out the lifestyles positively valued by the group in control of the government -negatively said, fewer people will carry out lifestyles or practices deemed as lacking in value or not valuable. The larger the interference of the state in civil society, the higher the scale and seriousness of the damage suffered by individuals in case of error. We may conclude that the seriousness and range of the damage caused in case of error will be higher in the case of a Perfectionist state than in a neutral liberal one.
Compliance with the justification requirements imposed by liberal neutrality results in an increase in the level of legitimacy of state actions. One of the features that characterize the culture of contemporary democracies is what Rawls called the fact of pluralism. People subscribe to conceptions of the good and religious, philosophical and moral comprehensive doctrines that are not only different from each other but, at times, are also irreconcilable. Let us presume that one group attains enough power to implement perfectionist policies that promote certain values or social practices associated, say, to their religious beliefs. Those subscribing to beliefs incompatible with the ones sustained by the majority -because they are atheists, or reject some of the central values of the religion at stake-will predictably consider that such policies lack justification. They may judge them based on unfounded or downright false beliefs. The situation is very different in the case of policies that satisfy neutral requirements of justification. Accepting the requirement of neutrality in justification implies a commitment to a process of political dialogue with a common language based on beliefs, values and forms of arguing and assessing evidence that are accessible to and accepted by all 23 so that everyone can acknowledge the weight of justification of principles or the actual policies ensuing from it. In view of our experience as citizens of contemporary democracies, we must admit that the hope of reaching a unanimous agreement is far from realistic, even after such deliberative process. Therefore it will ultimately be necessary to vote. 24 Even if no unanimous agreement is reached and it is necessary to vote, however, the situation after a public justification associated to the neutrality thesis will be very different from a situation where principles of justice or policies submitted to the decision of the majority express particular religious, philosophical or moral comprehensive doctrines. In this case, as we saw before, losers in the vote may consider that the winners' proposals are far from having an adequate justification. Contrariwise, if every policy submitted to a vote satisfies the requirements imposed by neutrality, those in the minority will have to acknowledge that the proposals supported by the majority are appropriately justified; despite the fact that they may continue to believe that their own point of view is better. This is so because the justification of those proposals will rest, at least on crucial issues, on beliefs to which they themselves subscribe and in ways of reasoning whose consistency they acknowledge.
We may reasonably presume that every subject will believe that the political conception she supports is more soundly justified than those of her opponents. She may think that it is a more adequate interpretation of the political ideals shared by the community, that it will be more efficient at promoting citizens' wellbeing, or that it is able to offer a stronger protection of the rights and freedoms of citizens, etc. The important point is that even in that case, provided the justification requirements set by the ideal of neutrality have been complied with, each agent must admit that the alternative proposals are justified by beliefs and reasons to which she had access. She cannot ignore their weight and soundness or, at a minimum, she cannot deny that they cannot foster the kind of rejection that might be caused if comprehensive doctrines incompatible with her own conception of the good were at stake. When citizens are committed to the type of deliberative process which, according to our exposition, is entailed by the ideal of state neutrality, the ensuing decisions, even if they happen to require a voting process, benefit from a degree of legitimacy incomparably higher that is the case with perfectionist policies. The increase in legitimacy of state policies entails a lower level of responsibility in case of error and corresponding harm to the parties affected. Stating that an agent has an adequate justification for her actions implies acknowledging that she cannot be held responsible 25 for the undesired results of her actions if she errs and fails to obtain the ends she was pursuing. We may conclude that, since a neutral state can attain a higher level of legitimacy than a perfectionist one, the degree of responsibility of the former if its policies should fail -that is, if instead of promoting citizens' wellbeing, they bring about the opposite resultwill be significantly lower.

195
In conclusion we could then state that a neutral state, as opposed to a perfectionist one, has the three advantages posited at the beginning of this section. It minimizes a) the possibility of error, b) the scale and seriousness of the damage it would cause and, c) the level of responsibility of the agents involved in the results in case it should actually occur.

1.3
It is convenient to pause briefly on the connection between justification and responsibility for the effects of actions. Let us presume that a doctor performs a surgery in the 19 th century. The surgery is successful; he removes a bullet from the patient's shoulder swiftly, with no harm to muscles or arteries. Despite that, the patient dies within a week because of an infection. Unknown at the time, the infection had been caused by bacteria in the doctor's hands and set of instruments. In fact, his action caused the death of the patient. Neither his contemporaries nor ourselves, however, would hold that the agent acted wrongly, qua doctor, or that he may be guilty of the ensuing result in either moral or legal terms. His actions were justified in view of the knowledge available and the standards of excellence at that point in time. There was no way the doctor could know that there were harmful bacteria in his hands and that they posed serious danger to the patient. Neither could he be aware of the procedure necessary to eliminate them. Needless to say, a contemporary doctor that failed to follow such procedure (say, if she performed a surgery without sterilised gloves on) would be fully responsible for the damage caused. This does not change our opinion on the doctor in the example. If we acknowledge that an agent can offer an adequate justification of her actions -and the justification will vary according to the context-no responsibility can be attributed to her for the consequences of those actions if they happened to be based on incorrect beliefs and on account of that they harmed third parties.
These considerations provide a further argument against perfectionism, based on the ideas developed in the previous sections. The dramatic increase in the seriousness of making an error upon acting in view of the fact that it is impossible to rule out the chance of such results requires to prioritise the courses of action that comply with at least one of the three features: a) minimise the probability of error, b) its scale and seriousness, or c) based on reasons not related to a and b, lower the level of responsibility to be attributed to the subject if the result does occur anyway. If an agent who can choose to act in a way that minimises (a), (b) or (c) decides to carry out an action with opposite characteristics she might be deemed fully responsible for the result of her action if she incurred in error. We have already seen that the passage from an action in a non-political context to a political one entails a dramatic increase -from a moral perspective-in the seriousness of incurring in error.
As opposed to perfectionists, neutral policies minimise the probability of error, its scale and seriousness and -for reasons unrelated to the factors mentioned-the level of responsibility of those concerned in case of error. If notwithstanding that, perfectionist policies were enacted, it seems clear that whether they fail and produce serious harm to those involved instead of promoting wellbeing, the agents who carried out that policies might be held responsible for the results. They could choose a course of action that would comply with more stringent standards of justification and they deliberately failed to do so. The situation is similar to the example of the train if we presume that despite being aware of the risk involved for the other person, the subject decided not to go into the trouble of checking whether it was in fact the express train. If the subject did that, she should acknowledge that if she actually made a mistake which resulted in the death of her friend she might be judged at least partially guilty of the outcome.
Acknowledging that in case of error we might be held responsible for the outcome of our actions is tantamount to an acknowledgement that we cannot provide an adequate justification to act along those lines. How do we come to that conclusion? Having an adequate justification for our action implies, as we said before, that even in the case of error we may not be held responsible for harm caused to third parties by our actions. On that premise, coming to the conclusion that one may be held responsible in case of error, is synonymous to acknowledging that one cannot provide such justification. We may conclude that perfectionist policies are always a legitimate object of reasonable rejection and that, upon enacting them, the state fails to honour its basic commitment to provide a justification of its essential policies that may prove acceptable to all parties concerned.
When renouncing to the implementation of perfectionist public policies the agents give up what they regard as the bigger benefits.

196
The proposed argument aims at expounding the reasons why this must be done. As we have seen, the context of action of democratic politics has special features -the scale of harm in case of mistake, the level of responsibility of the agents in case of mistake, etc. This entails that a small, though significant, chance of mistake is sufficient to have good reasons to refrain from acting or to be compelled to look for public policies meeting more demanding standards of justification. This argument discloses the fact that the agents must acknowledge that they know beforehand that they are not justified to act on the basis of their comprehensive doctrines. This is due to the fact that they know that if they implement the perfectionist public policies they support, they would be responsible for harming others in case of mistake. Being justified to act implies that in case of mistake our responsibility for the resulting harm is minimized or neutralized. Knowing that in a certain context this does not obtain entails that we are not justified to act, which is not modified by the fact that we may expect a big good in case things go well. Imagine that a medical doctor identifies two alternative treatments to deal with the illness of one of her patients. One of these could cause great harm to the patient in case of mistake, but may result in a complete recover if things go well. There are high chances that things will go well, still one cannot be completely sure that the treatment will not fail. The other treatment will not bring about a complete recover but it is harmless or much less harmful in case of mistake. The patient chooses to follow the latter treatment. Now suppose that the doctor decides to force the patient to follow the former, say, by cheating her. We are aware that the doctor is not justified to do this even though this may increase the chances of improving the patient's quality of life. We know that the doctor is not justified to act in such a way, at least in part, because we know that in case the treatment fails, the doctor will be responsible for the resulting harm even if the harm was not caused by any fault on her part. When we decide on the implementation of perfectionist public policies we are in a similar scenario.
I want to briefly consider some potential objections. Let us suppose that advocates of perfectionism claim that they are capable of reaching the more stringent standards of justification for acting that are required to move from non-political to political contexts -I have referred to this as moving from C1 to C2. Their argument is simple.
They maintain that they have examined their comprehensive religious, philosophical or moral doctrines in a more careful way, or that that they have discussed their beliefs with people subscribing to other doctrines, and, after this deliberative process, they are even more convinced of the truth of their viewpoints. They have conduced further research and have thus confirmed their beliefs. Why would not this be enough? As already argued, in the train case, gathering more information confirming that the train in the station is the express one may be sufficient to justify S2 to act. The problem is that, as said before, political contexts do not follow this pattern. These cases are more similar to that of the scientists, introduced at the end of section I. In that case, increasing the level of epistemic justification of T1 was not enough to be justified to act. In the case of the scientists, the reason why B was not justified to act relied on his acknowledgement of three factors: 1) that the level of harm that making a mistake in his research would cause to third parties would be enormous; 2) that the chance of being mistaken was always significant, however convinced he was; and 3) that an important weight must be attributed to the potential disagreement of those affected by his activities. Gathering more information in support of T1 neither cancels nor diminishes any of these factors. Now, the political case is somehow similar to this. Whoever aims at modeling the institutions of the basic structure of his society according to a comprehensive doctrine, must acknowledge that the following features obtain for him: 1) the chance of making a mistake, however small, is never trivial or insignificant; 2) the harm that he would cause to the potential victims would be grave, perhaps irreversible; 3) many of the people that would be affected by her activities starkly reject the public policies he defends and deny any validity to the premises supporting them. The acting agent is also aware that these factors can be neutralized by reformulating his proposals in such a way that they meet neutral standards of justification, that is, proposals that are grounded on beliefs that are acceptable to all the affected parties. The agent knows, ultimately, that he is not justified to act in that sphere on the grounds of his comprehensive doctrine, since he must acknowledge that if he is mistaken he would be responsible for the harm caused to his fellow citizens. Instead, if he satisfies neutral standards of justification, both the harm and his responsibility for that harm, will be neutralized or severely diminished. 26 The fact that the advocate of perfectionism claims that he has further deliberated or researched, which constitutes a cognitive gain, and that he is therefore more convinced of the truth of his comprehensive doctrine, does not cancel any of the premises of the proposed argument. The chance of mistake as well as the other above mentioned factors remains. This type of criticism would only work in a sort of cases that I have excluded from the very beginning, namely, those cases in which the agent's cognitive progress allowed him to know that his doctrine is true in an infallible or incorrigible way. 27 There are other two potential objections I would like to consider. The first of these claims that exactly the same reasons that my argument invokes against perfectionism can be used against a liberal state. My argument would then lead to a minimal state such as that promoted by some libertarian theorists. The reasoning underlying this conclusion is simple. If the main advantage of a liberal state over a perfectionist one is that it minimizes the chances of mistake and the scope of the harm in case of mistake, a libertarian state has this same advantage over a liberal state. As long as a libertarian state would have less influence on social life than a liberal state -which may include, for instance, the implementation of measures addressed at redistributing wealth-a libertarian state would be less harmful than a liberal state in case of actual mistake. The second objection maintains, on the other hand, that my argument fails in excluding some moderate variants of perfectionism. If a state simply promotes certain values or activities in a non coercive way, its influence on social life will be minor than that of a state applying coactive measures. At the same time, since individuals are free to accept or reject any incentives the state may offer them to influence their lives in certain ways, they will be responsible of their own choices. In case something goes wrong, it will be misleading to claim that responsibility falls exclusively on the state -as may indeed be claimed in the case of coercitive perfectionist policies. Now I want first to point out that these objections contradict each other. Thus, if any of these was correct, one must conclude that the other one is mistaken. If my argument leads to a minimal state, then it excludes the moderate variants of perfectionism. If, instead, it is indeed compatible with some moderate variants of perfectionism, it cannot be accused of leading to a libertarian approach. I think, nevertheless, that both objections are misleading for similar reasons. My argument singles out three equally important advantages that neutral standards of justification have over any perfectionist justification of public policies. As has already been said, when compared to perfectionist public policies, neutral public policies minimize a) the probability of making a mistake, b) the magnitude of the harm that can be caused in case of mistake, and c) increase the legitimacy of public policies by offering a justification of these based on beliefs and ways of reasoning that anyone can accept. Although these objections contradict each other, it seems that both steam from not taking duly into account the role that variable c) plays in my argument. The moral consequences of lack of legitimacy are really relevant. It often happens that public policies considered harmless at first sight become harmful when coupled with adverse circumstances. For example, devoting resources to the implementation of perfectionist public policies -either by coercive or non coercive means-may render the state incapable of facing an unexpected economic crisis. The perfectionist policies which costs bring about this state of affairs will not be justified for those citizens that reject the comprehensive doctrines that underpin them. The state will be fully responsible for any harm that these people may suffer as a result of the economic crisis. This is why even the most moderate perfectionist public policies are to be avoided. Now, in case it was impossible to meet the ideal of political legitimacy introduced in c), then, my reasoning would lead to a libertarian position. However, this is not the case. There is no reason the think that such an ideal of political legitimacy cannot be reasonably achieved. A deliberative regime where the citizens proposed several political conceptions of justice could fulfill it. 28 This should vanish any suspicion that my argument may lead to some sort of libertarianism or even anarchism. In so far as a proposal is better justified, the reasons for minimizing the risk of mistake and its possible consequences become less stringent. This explains that we may be allowed to undertake considerable risks by applying public policies aimed at improving the life prospects of the people within the domain of a neutral justification grounded on common beliefs -the kind of egalitarian policies that can be deduced from Ralws' political liberalism are an example of this. Therefore, it is perfectly possible to justify public policies that go far beyond libertarianism by reference to a political conception of justice; there is no reason to reject these policies insofar as the criterion of legitimacy obtains. 29 1 According to Rawls' definition, a comprehensive doctrine includes "conceptions of what is of value in human life, and ideals of personal character, as well as ideals of friendship and of familial and associational relationships, and much else that is to inform our conduct, and in the limit to our life as a whole" (Rawls, 1993, p. 13). 2 Cf. Rawls, 1993, pp. 134-135;Chan, 2000, pp. 5-6;andRaz, 1989, pp. 1230-32. 3 The case is a variation of the example of the airport, originally proposed by Stewart Cohen (cf. Cohen, 1999), and similar to the one used by Fantl and McGrath in "Evidence, Pragmatics, and Justification" (cf. Fantl and McGrath, 2002). 4 Cohen, 1999, p. 58. 5 Cf. Fantl and McGrath, 2002, and 2007 McGrath are subject-sensitive invariantists (SSI). This view attempts to capture the intuitions of the contextualist without claiming that "know" is an indexical expression. Against contextualism, they maintain that "know" may change its content according to the context. Another difference is that for SSI the knowledge relation is sensitive to what is at stake for the subject rather than what is in question for the agent who makes the knowledge attribution. For a detailed comparison between both positions, see Timothy Williamson, 2005, pp. 214-220. 6 The authors use an independent argument to back their stance. 7 Other structurally similar cases to that of the train have been used in various discussions. Besides those mentioned above, we may add the following examples. In an article published in the 50's, Richard Rudner refers to differences in the cost of error when testing scientific theories in order to back his thesis according to which value judgements are not external to scientific practice since they would affect the very process of theory-justification (cf. Rudner, 1953). An additional recent example is DeRose's defence of epistemological contextualism (cf. DeRose, 1992). Mark Kaplan offers an interesting Bayesian interpretation of the implications of these cases on the concept of beliefs (cf. Kaplan, 1996, pp. 105-6). The impact of variations in the cost of error plays a relevant role in the distinction forwarded by Richard Foley between epistemically rational belief and justified belief (cf. Foley, 2005, pp. 313-326). This sort of consideration also plays an important role in the distinction between belief and acceptance (cf. Bratman, 1992 andStalnaker, 1984, pp. 79 ff) and in the draft of a contextualist concept of the notion of belief proposed by Robert Nozick (cf. Nozick, 1993, pp. 96-97). 8 It is worth stressing that the argument I propose hereby does not rely on any sort of commitment with contextualism or subject-sensitive invariantism. The proposed argument departs from certain intuitions that play an important role on that views, but conceptualize them in a different way. This is the thesis that one may be justified in believing that p in a certain context without being, however, justified to act on the ground of p in other context, notable in those contexts where the making a mistake is more dangerous. This distinction between being justified to believe and being justified to act is not suggested by any of these epistemological views. 9 For example, S1 may have heard someone remark that the previous day the information board at the station was not working properly. 10 Gaus, 1995, p. 242. 11 Though Gaus' context of discussion is very different from that posited here, he reaches similar conclusions to those I am defending as to the difference between being justified in believing and acting and the fact that the possibility of refraining from action may be compatible with a firm belief which -however-is set aside when acting within a specific context. (cf. Gaus, 1995, pp. 241-242). 12 Richard Foley brings in a similar example to hold that social roles may have an impact on the degree of stringency to be satisfied by the standards required to judge that an agent's beliefs are justified. According to Foley, "The concept of justified belief is also able to give expression to the way in which in our everyday assessments of each other's beliefs, the intellectual standards we expect one to meet vary not only with the importance of the topic at issue but also with one's social role. If it is your job but not mine to keep safety equipment in good working order, the intellectual demands upon you to have accurate beliefs about the equipment are more stringent than those upon me. My belief that the equipment is in good working order might be justified even if I have done little, if any, investigation on the matter. I need not have tested the equipment, for example. A cursory look might suffice for me, but this won't do for you.
[…] The standards of justified belief are higher for you. You need to do more, and know more, than I in order to have a justified belief about this matter" (Foley, 2005, p. 324). Even though the stance I am defending here is more modest, it shows a few similarities with some of Foley's ideas. One of the central issues in his proposal gears around the distinction between justified belief and epistemically rational belief. Practical matters -such as the cost of error or the level of special responsibility associated to some social functions or professions-would impact, albeit indirectly, on the first concept. Consequently, standards to be complied with by a belief in order to qualify as a justified belief would at times be lower and at other times higher than those that apply to an epistemically rational belief. From Foley's point of view, in various ways our "non-epistemic ends" and the sort of practical factors associated with them that we have considered "help to determine what one can justifiably believe [….]. The idea is not that they give good reasons to believe a proposition for which one lacks good evidence. Rather, they define the extent of evidence gathering and processing that it is reasonable to engage in with respect to a particular issue. They thus shape what it is justified for one to believe in an indirect way rather than a direct [one]" (Foley, 2005, p. 324). Contrary to Foley, I refrain from passing judgment on the issue whether practical matters and our non-epistemic ends shape or do not shape our beliefs. I only state that those matters play a crucial role when evaluating issues that concern the justification of actions. I do not pass judgment on whether a social function -such as being in charge of the technical maintenance of a potentially dangerous mechanism-should have an impact on the standards of justification that turn the agent into a "responsible believer". I am only saying that they hold a decisive influence in the evaluation of the subject as a "responsible agent", that is, when assessing if her decisions and actions are justified or otherwise. 13 As you may have noticed, in this case satisfying more stringent standards of justification for acting, does not entail satisfying standards more stringent from an epistemic point of view. None of the factors that the satisfaction of which would allow B to meet the accurate standards of justification, namely, minimization of harm, or agreement on the part of those affected by the experiment, has any cognitive upshot. Both factors are independent from the probability of T1 being true. 14 Or they do not occur with comparable intensity. 15 Cf. Rawls, 1993, p. 227. 16 This does not mean, as Rawls says, that in effect political power is "always coercive power", which seems clearly wrong as states frequently apply policies aimed at directing the behaviour of individuals through economic and other kinds of incentives, with no sanctions imposed on those who do not make the decisions which are being promoted. For criticism of Rawls' stance on this, see Kymlicka, 1996. 17 Nagel, 2005 Perfectionists naturally acknowledge this duty. Their idea is that as their conception of the good or comprehensive doctrine is correct, its political implementation will impartially promote the good of all affected parties and, consequently, will be seen as adequately justified by all citizens. 19 As we know, these principles state that "a.. Each person has an equal claim to a fully adequate scheme of equal basic rights and liberties, which scheme is compatible with the same scheme for all; and in this scheme the equal political liberties, and only those liberties, are to be guaranteed their fair value" and "b. Social and economic inequalities are to satisfy two conditions: first, they are to be attached to positions and offices open to all under conditions of fair equality of opportunity; and second, they are to be to the greatest benefit of the least advantaged members of society" (Rawls, 1993, pp. 5-6) 20 Cf. Rawls, 2001, pp. 26-27. 21 Cf. Rawls, 1999a For an influent development of this idea see Popper, 1965, pp. 217-220. 23 Rawls claims that "in discussing constitutional essentials and matters of basic justice we are not to appeal to comprehensive religious and philosophical doc-trines [...] As far as possible, the knowledge and ways of reasoning that ground our affirming the principles of justice and their application to constitutional essentials and basic justice are to rest on the plain truths now widely accepted, or available, to citizens generally. Otherwise, the political conception would not provide a public basis of justification" (Rawls, 1993, pp. 224-225). 24 Rawls is of the same opinion on this point. As we know, he holds that satisfying the demands of the liberal conception of legitimacy implies offering a justification of basic State policies through a political conception of justice. Notwithstanding that, he is aware of the fact that political debate in a democracy will almost inexorably lead to the formulation of various reasonable political conceptions of justice that will have to compete against each other in elections. In his opinion, within contemporary democracy "The content of public reason is given by a family of political conceptions of justice, and not by a single one. There are many liberalisms and related views, and therefore many forms of public reason specified by a family of reasonable political conceptions" (Rawls, 1999, pp. 140-141). According to Rawls, "when hotly disputed questions […] arise which may lead to a stand-off between different political conceptions, citizens must vote on the question according to their complete ordering of political values. Indeed, this is a normal case: unanimity of views is not to be expected. Reasonable political conceptions of justice do not always lead to the same conclusions, nor do citizens holding the same conception always agree on particular issues" (Rawls, 1999, p. 169). 25 Or at least, her degree of responsibility diminishes significantly. 26 As we have seen, the adoption of neutral standards of justification entails, apart from the moral advantages already mentioned, an epistemic advantage, namely: the reduction of the possibility of making mistakes. 27 Throughout this essay I have assumed that my addressee is a reasonable perfectionist, and that a necessary condition for being reasonable consists in recognizing that we cannot achieve a complete, definitive knowledge in empirical or valorative matters. 28 By definition, such conceptions would be insufficient as a basis for justifying perfectionist public policies. 29 A clarification regarding the scope of my argument is in order. This argument has been designed to be applied to a modern constitutional democracy characterized by deep and irreconcilable disagreements. If, say, the members of a certain society unanimously shared an agreement on some comprehensive doctrine they could achieve, when justifying their public policies, a level of political legitimacy similar to that reached by a pluralist society by the means of a neutral justification. In such a situation the reasons I have invoked for rejecting perfectionism would be in neutralized, or almost neutralized. But the point I wish to stress is that the argument propose here does not claim any validity in such a context: as Ralws' political liberalism, it has been designed to apply only to societies characterized by the fact of pluralism.