Types of deliberation

As research has increasingly addressed deliberative processes through theoretical analyses, empirical studies, and practical experiments, it has become apparent that deliberation is a phenomenon with many faces. Argument-based interaction may come about in different ways and have different functions. Many of these differences can be explained by the nature of the entry positions, i.e. the type of judgment participants express at the start of the deliberation process. Positions may be strong or weak, conscious or unconscious, free or constrained. I discuss the relationship between type of position and type of deliberation, hypothesizing what difference there may be in the development of the deliberative process and in its outcome, and look at the most frequent deviations from the ideal deliberative model. I discuss both cases of symmetrical deliberation, in which all participants express positions of the same type, and cases of asymmetrical deliberation, more frequent in the real world, in which interaction is among actors whose positions are of different types. The analysis suggests that specific settings should be adopted and specific strategies employed depending on the type of deliberation involved.

deliberative aspects in the interaction they observe among participants.It has become apparent that collective decisions are not only the fruit of negotiation or of preference aggregation, but are also the outcome of dialogic processes in which participants' opinions are shaped or modified.Deliberative practices are systematically observed in a wide range of situations: in parliaments (Bessette, 1994;Steiner et al., 2005), in European policy making (Eriksen and Fossum, 2000;Naurin, 2007), in local development policies (Cersosimo and Wolleb, 2006), within social movements (Polletta, 2002;della Porta, 2005) and in many other fields.Nor should we leave out the numerous experiments of artificial forums -such as citizens' juries, deliberative polling, the 21 st Century Town Meetings, the Canadian Citizens' Assemblies on electoral reform, the Australian Citizens' Parliament -that have been designed precisely to offer a favorable surrounding for deliberation (Fishkin, 1995;Fung, 2003;Gastil and Levine, 2005;Warren and Pearse, 2008;Carson, 2008;Goodin, 2008;Dryzek, 2009).Wherever a decision-making process among different viewpoints comes about, some trace of deliberation may be observed.This across-the-board use of the concept, however, brings to light a doubt: when we speak of deliberation, are we talking about one and the same thing?Might we not be looking at a multiplicity of practices that, while they share the use of argument in dialogic form, for the rest present contrasting aspects?After all, why should a parliamentary discussion resemble a meeting of the No Global movement, or a stakeholder partnership, or a citizens' jury?Should we not therefore conclude that: 'deliberation cannot be understood as a unified process' and that 'to speak of a unified thing called deliberation is to speak of a chimera' (Button and Mattson, 1999, p. 619)?
We may reason along similar lines in connection with the drawbacks or pathologies (Hamlett and Cobb, 2007) of deliberation.The dialogic exchange should have the effect of bringing participants' viewpoints closer together; it should produce better decisions, and should reinforce their legitimacy.But the results of deliberation are frequently in contrast with those ambitions.Discussion can generate conformism, polarization (Sunstein, 2002), positional contrast; it may be subject to manipulation, it may favor those persons who master rational argumentation, to the detriment of those who are accustomed to express themselves in narrative form (Sanders, 1997); it may exclude rather than including (Young, 2000); it may exacerbate conflicts rather than settling them; it may induce individuals to strengthen their initial beliefs rather than to re-examine them in the light of other participants' arguments (Manin, 2005).Here, too, we must ask whether these faults are faults of deliberation as such, or whether they are likely to appear in specific deliberative contexts.It is far from certain that all deliberative processes run the same risks.Some of them are more likely than others to degenerate in specific ways.
But how can we distinguish among the different deliberative processes?The hypothesis that I aim to discuss in this article is that the development of deliberation depends on the participants' different 'entry positions', i.e. the nature of the judgments they express at the start of the deliberative process.If deliberation aims to foster a mutual understanding among actors' viewpoints, it is legitimate to suppose that the process will come about in different ways according to the type of their 'predeliberative opinions' (Barabas, 2004)  1 .
Each participant arrives at the dialogic forum with his or her own judgment on the issue that is under discussion.These initial positions depend on preferences, on beliefs 1 When talking about 'entry positions' it is clear that I am referring to discrete deliberative sessions in delimited places and times.In the literature these situations are normally defined as 'micro deliberation'.My analysis does not concern 'macro deliberation', i.e. the ongoing process of deliberation that spontaneously occurs within society (Hendricks, 2006;Parkinson, 2006;Goodin, 2008) (I thank the anonymous reviewer for this comment).In fact my main concern is about the design of such micro processes.concerning the state of the world and cause-effect relationships; they must be accompanied by arguments that, at least presumably, are capable of being justified in the eyes of the public.The point is that the nature of these initial positions may vary greatly across different situations.They may be more or less definite, more or less solid, more or less malleable.Not all participants enter the deliberative process with equally well structured or equally firm convictions.And this initial aspect is likely to influence the following process.
Traces of this line of research may be found in some empirical studies that compared deliberative forums among 'professional', 'partisans', 'insiders' on one side and those among 'non partisans', 'lay people', 'ordinary citizens' on the other.Such studies repeatedly showed that discussion develops differently in the two contexts and tends to bring to different outcomes (Petts, 2002;von Stokkom, 2005;Jackman and Sniderman, 2006;Hendriks et al., 2007).Fung (2003) maintained that what does change is the 'temperature' of the deliberative process which in 'hot' in the former cases and 'cold' in the latter ones.In a comparative analysis of seven forums held in the United States, Button and Mattson observed that models of deliberative democracy 'assume pre-existing and ordered preferences among citizens … [while in our cases participants] came to their opinions about the issues being discussed as a result of the process of talking and listening to others' (Button and Mattson, 1999, p. 621).Thus the deliberation ended up to have an educational goal that 'locked citizens into a deferential and sometimes passive role.When political learning is the focus, citizens become pupils rather than participants' (ivi p. 622).Steiner et al. (2005) note the opposite phenomenon in their study on parliamentary debates: here the educational function of deliberation is virtually missing; participants appear to be closed on their initial positions.But, the authors add, 'this may not hold true in some other settings such as civic forums, where partisan divides are often less relevant and initial positions may not be crystallized' (Steiner et al., 2005, p. 136).

Types of position
The clues that emerge from these studies suggest that deliberation takes on different features depending on whether the dialogue comes about among insiders (experts, politicians, bureaucrats, stakeholders, representatives of interest groups) or among lay citizens.It appears that the nature of participants' starting positions can affect both the 'temperature' of deliberation and its effectiveness (and possibly other aspects too).We may proceed along this path by examining how the initial positions may be different, and which differences may be relevant to the development of the deliberative process and to any fault it might suffer from.
The positions participants take at the start of the deliberative process may vary along three dimensions.The first dimension is the degree of definition of those positions.
He or she who enters a deliberative arena may have already developed firm convictions on the issue under discussion, or may be in a condition of doubt or uncertainty.In the public sphere any citizen may, sooner or later, be faced with a theme on which he or she has vague and confused ideas or, simply, has no particular opinion.Frequently, in consequence, his or her position is also unstable: the person may vacillate in the space of a short time between very distant viewpoints depending on the stimuli to which he or she is exposed.Thus we may consider as well defined positions those positions that are structured, stable, with few doubts; and, vice versa, ill-defined positions are those that are loose, unstable, uncertain or surrounded by doubts.
The second dimension is the reflectiveness of participants' initial judgments.A person may express a more or less reflective opinion depending on the information he or she has available on the issue.Information that is relevant to formulating a position is not limited simply to facts concerning the problem under discussion.Equally important, if not more so, is the reflection about the implications that an option may involve, for oneself, one's community or for others, within a shorter or longer timeframe, as well as a knowledge of the state of the debate on the issue: which arguments for and against are raised, and by whom.Thus we may consider as reflective those positions that take into account opposing arguments that are raised in debate, and that are capable of anticipating the consequences that the preferred solution might entail.Vice versa, we may consider as unreflective the positions of those who are not aware of the arguments supporting different options from their own, and who do not realize the implications that their own options would involve.It must be added that it is obviously of great account whether the participants' judgments are free or whether they are under some external constraint.Frequently, in theoretical accounts of deliberation, it is assumed that participants only answer to themselves and are thus free to change their points of view in the light of the content of the discussion.But this is often not the case: those who participate in a debate may be called upon to respond for their opinions to others (a party, an association, a constituency, etc.) and their choices may be subjected to some type of ratification.
This third dimension (freedom of positions) is not entirely independent of the other two (degree of definition and reflectiveness).In the situations I have called 'suspension of judgment', 'uncertainty of judgment' and 'prejudice', we can suppose that the participants are free to change their minds.On the contrary it is much more probable that some type of constraint exists in the situation I have call 'certainty of judgment'.Participants who express well defined and reflective positions are chiefly spokespersons of organized groups.Even were they convinced of the good reasons put forth by other participants; they would not always be free to declare that fact publicly.The force of the constraints may vary depending on the circumstances or on the theme under discussion, but it is unlikely to be completely missing.Thus from now on I will need to take into account that positions of the first type are not only 'well defined and reflective', but are also -to some extentsubject to some external constraint.
It is worth noting that the nature of initial positions of participants is not determined only by the psychological traits of their personality.Of course psychology matters: people may be more or less mind-opened or conformist as shown by the studies on the psychological dynamics of deliberation (Reykowski, 2006;Mannarini, 2009).On the other hand, also roles matter: people accustomed to act in the public sphere as spokepersons of parties or interest groups tend to stick to fixed convictions.But, to some extent, the nature of the positions is also determined by the characteristics of the context within which the deliberation takes place.As we shall see, one person may have positions more or less malleable, more or less reflective according to the context in which deliberation occurs.

Symmetrical and asymmetrical deliberation
It is very probable that deliberative forums that take place in the real world will bring together people who express positions of different types.It is likely that, in a meeting, in a committee, in a forum, there will be a mixture of people expressing judgments that are well and ill-defined, reflective and unreflective, free and constrained.
Thus we may say that deliberative processes in the main present an asymmetrical nature, since they take place between people who are not equal with regard to the strength of their convictions, their awareness of the problem, or their freedom to reformulate their viewpoint during the discussion.
Nevertheless, it is not impossible in the real world to find cases of symmetrical deliberation, i.e. specific spheres in which the debate takes place among participants who express positions of the same type.We will see later what these cases are.But leaving aside the empirical evidence, situations of symmetrical deliberation are of great interest from the analytical standpoint, because they enable us to study some ideal types of deliberation in what we might call the pure state.

Models of symmetrical deliberation
From the analytical standpoint (but, as we will see, also in empirical terms, at least to some extent) we can distinguish four types of symmetrical deliberation, which It might seem improbable that a deliberative forum be made up exclusively of individuals with these characteristics (which are not common ones).However, since some deliberative settings do exist that encourage participants to discuss on the basis of a certain epoché, accepted and shared by all, situations of this type are not at all impossible.An example may be that of committees of experts who are called upon to define a standard or to give an evaluation for a public decision.A similar situation could be perhaps found in parliamentary commissions, when the theme under debate is not too hot and when the divisions among participants do not correspond to party cleavages.However, the classic example, from which much of the theory of deliberation began, consists of juries in criminal trials (Gastil et al., 2002;Gastil and Weiser, 2006).The fact that juries are drawn by lot reduces the likelihood of predetermined positions.And once the deliberative phase itself begins all participants have a deep and absolutely identical knowledge of the case under discussion.They have listened to the witnesses and to the opposing arguments brought by prosecution and defense, and thus they are in the ideal condition to find a solution through dialogue.They may not succeed in reaching agreement, but the path they follow is likely to be based on reciprocal listening and common reflection.
The account proposed by Gustavo Zagrebelsky (2005) on how the Italian Constitutional Court works corresponds in full to this model.When the Court meets in the counsel chamber -reports Zagrebelsky -the judge initially tend to take positions and sides, but as the discussion continues 'the objective need to reach agreement takes pride of place….There is a natural tendency to understand the reasons of others' (ivi, p. 43).The very deliberative context of the Court pushes the judges to take an open and dialogic approach: in the Court 'everything possible is done not to vote or, better, to deliberate without it being necessary to turn to voting, or to make it into a simple formality.This requires commitment and prolonged discussion, which could easily be avoided by a simple deciding vote.On the contrary it is Court wisdom to resist the temptation and give itself the necessary time, without forcing things' (ivi, p. 42).'The deciding vote is extrema ratio… But when the Court votes there is always a sense of bitterness even among those who are in the majority' (ivi, pp. 47-48).
The symmetrical deliberation in conditions of suspended judgment is that which comes closest to the ideal speech situation proposed by Habermas (1987).Here, indeed, the best argument has the greatest probability of winning over participants.We would be tempted to say that the scope of theories of deliberation (like that of Habermas) is not as general as is frequently supposed.Rather they tend to be applicable to the specific circumstance described here, that is when participants express ill-defined but reflective judgments.In other cases -as we will see -the possibility of reproducing an ideal speech situation is much more remote.
Another characteristic of this model of deliberation is its dual nature: it is public and not public.Deliberative forums of this type (juries, constitutional courts, committees of experts, perhaps also parliamentary committees) are public arenas in which participants are induced to formulate impartial arguments of general scope.But at the same time they are circumscribed public spaces that are clearly separated from the general public.The deliberative process takes place sheltered from the eyes of the world and this discourages participants from taking demagogic positions (Elster, 1998b;Stasavage, 2007).This balance between 'public' and 'not public' is delicate and precarious; but it is one of the most important institutional guarantees for the success of the deliberative process (Chambers, 2004).
This does not mean that, even in this specific context, there is no risk of deliberation's degenerating, but these risks are more limited than in any other context.The fact that participants' initial positions are not fully pre-defined reduces the risk of their taking defensive positions and lessens positional confrontation; participants' reflectiveness makes the effects of conformism, polarization and manipulation less probable.
These deliberative situations are extremely rare.However, we may ask whether it might be possible to reproduce in other contexts the particular features of this setting that induce participants to take attitudes favorable to dialogue.Or, in other words, whether it is possible to determine institutional strategies that can make experiences of this type less rare.Not all these features would be easy to reproduce; for example, deliberation appears to be favored when a common language or common knowledge is shared by participants, even if there is disagreement about the subject matter.Legal expertise can facilitate interaction among constitutional judges, as can the technical and scientific knowledge shared by committees of experts (note, though, that this shared medium is completely missing from juries).However, it would be very difficult to artificially reproduce this condition in other deliberative contexts.
Other aspects lend themselves more easily to transposition: setting up public spaces that are circumscribed and not open to the public; providing complete information that is equal for all; giving time to discuss and to reach agreement.

Type b: Certainty of judgment
In the second model of symmetrical deliberation, the participants are well-informed and know exactly which side they are on.Unlike the previous case they have few doubts.
They have positions to defend and strategies to follow.Frequently they must report back to external bodies and thus are not completely free to change their entry positions.This situation typically occurs when deliberation takes place between stakeholders, activists, militants, party or interest group representatives.Situations characterized by 'certainty of judgment' are much more frequent than the previous ones.Parliamentary debates are likely to be of this type, as well as the numerous stakeholders' partnerships.
In these settings the deliberative process meets with greater difficulty.Deliberation would aim at softening the participants' original positions and making them slightly more compatible, but this possibility is limited by the rigidity of initial positions.Some degenerative aspects are highly improbable: the fact that participants have well defined and reflective convictions puts a brake on the risk both of polarization toward extreme positions and of manipulation.
Two types of distortion of the deliberative process are, on the contrary, highly probable, and tend to produce opposing effects.First and foremost, the dialogue may harden to become pure and simple positional confrontation in which the parties leave the discussion with exactly the same opinions with which they began.It is hardly surprising that, in their empirical research on parliamentary debates, Jürg Steiner and collaborators found that speeches tending to put forward and stress one's own position are overwhelmingly preponderant (they appear in 82 percent of 2,995 speech acts analyzed), whereas formulation of alternative proposals, appeals to consensus, and proposals for mediation play marginal roles (respectively only 8 percent, 7 percent and 3 percent of speech acts) (Steiner et al., 2005, in particular p. 132).The tendency towards a 'dialogue of the deaf' may mean that debate turns into a ritual display, aimed more at external constituencies than at any attempt to find some common ground.
This type of symmetrical deliberation, however, also runs the opposite risk: that of sliding into bargaining.When positions are distant and the discussion is unable to bring them closer, it is often possible to resolve the contrast through mutual concessions and thus to reach an agreement that does not result from a common view.Rather it is a settlement that leaves participants' original points of view substantially intact.Negotiating tables for local development -just to given an example -can be resolved through a distributive agreement that enables advantages to be allocated to participants, without creating any common good (Magnatti et al,. 2005;Cersosimo and Wolleb, 2006) 2 .
Nevertheless, positional confrontation and negotiation are far from being the only possible outcome.Empirical research has shown cases of symmetrical deliberation among insiders that succeed in overcoming the distance between their initial positions and in reaching, through discussion, new common goals.This is what happens to the French mayors studied by Pinson (2005); by interacting in inter-municipal bodies they succeed in reformulating problems and finding innovative solutions for their communities.Or, in Italy, those virtuous cases of agreement over local development (Magnatti et al., 2005;Cersosimo and Wolleb, 2006) in which, through discussion, participants succeed in reworking their points of view and in producing a common vision that none of them possessed initially.
It is not clear which institutional aspects tend to push the participants in the direction of positional contraposition or negotiation on one hand, or towards deliberative practices on the other.It might be thought that different status of publicity play a role.
Parliamentary debates (especially if they are in the House) take place in highly public conditions and are thus more exposed to positional confrontation.Discussions at negotiating tables or in meetings of the communautés d'agglomération take place sheltered from outside eyes and thus a less rigid confrontation is favored.Nevertheless, a high degree of secrecy combined with a limited number of participants is the institutional formula most favorable to finding settlements of the negotiated type (Elster, 1998b).
Deliberation appears to require the existence of a public space that is not too exposed to the public.
2 Negotiation can be understood as a legitimate way to overcome hard differences even for deliberative democrats , if no coercive mechanisms are at place (Mansbridge, 2010), but it can bring to distributive outcomes that simply 'divide the cake' among participants without any benefit for the broader community.
Another clue is offered by recent analyses on the role of leadership in negotiating tables for local development (Magnatti et al., 2005;Barbera, 2005;Cersosimo and Wolleb, 2006).According to these studies, the existence of a leader tends to favor the development of an effective deliberative process, not so much, or not only, thanks to his or her personal authority or charisma, but above all because he or she is capable of showing the participants common goals that they can share, while at the same time renouncing the possibility of obtaining specific advantages for him or herself.It appears that, for the process to result in dialogue, a figure is required who is capable of standing above the interests at stake, and who is able to provide mediation services or to facilitate interactions.
It might be concluded that symmetrical deliberation among insiders could benefit (even if we do not know how much) from the right condition of publicity and from the presence of mediators or facilitators who are outside the fray; they should be capable of providing structure to the decision-making process and of helping participants to perceive collective goals that, alone, they would not have been able to see.Politicians, militants and activists, in a word all those who express well defined and reflective positions, are in general not very willing to be helped.They prefer to go it alone.In the next two cases we will see that, on the contrary, support for decision-making is usual, indeed absolutely necessary.And we may perhaps be able to find some suggestions that can also apply to the deliberative contexts we have already looked at.

Type c: Uncertainty of judgment
We will now turn to the opposite situation case of symmetrical deliberation in which participants are not fully aware of the stake and have poorly-defined viewpoints.
Here, uncertainty of judgment holds sway: people have vague and confused ideas, they have available only scraps of information that is often random and uncoordinated, they may have preferences concerning some aspects of the issue but they hardly see the wider picture.Frequently their options are contradictory because they tend to embrace different points of view depending on the specific aspect that is focused.
In the real world it is highly improbable that people with positions of this type (illdefined and unreflective) decide to meet for the purpose of taking collective decisions.
Forums of the 'uncertainty of judgment' type are generally the fruit of experiments that are artificially constructed to give a voice to ordinary citizens.Sometimes participation is voluntary, as is the case of the American National Issues Forum (Melville et al., 2005) and others (Button and Mattson, 1999;Fung, 2003).Sometimes the participants are randomly selected from among the population involved in the issue under discussion, as comes about with citizens' juries (Smith and Wales, 2000;Crosby and Nethercut, 2005), consensus conferences (Hendriks, 2005b), deliberative pollings (Fishkin, 1995) and many other similar experiences (Fung, 2003;Gastil and Levine, 2005;Warren andPearse, 2008, Dryzek 2009).
In these cases, the deliberative process takes place very differently than in the two contexts considered above.Here, help with the deliberation process is a necessity: the participants are not left to themselves but enter a pre-structured process.Generally the experimental design is drawn up by an advisory committee, comprising all the main stakeholders, which guarantees to provide balanced information.There is a rigid timeframe for the proceedings.Facilitators assist discussion among the participants.The fact that the process is structured does not compromise its informality: on the contrary, the organizational framework is set up to stimulate the development of informal relations and to eliminate those barriers that are frequently present in non-structured interactions.
But what changes in the deliberative process is not only its form, bit also its function.While in situations of 'certainty of judgment' and in those of 'suspension of judgment', the goal of deliberation is to bring participants' positions closer, here the goal is first and foremost to shape those positions, to enable participants to gain an idea.
Information is offered during the process.Participants' positions do not constitute the input into the deliberative process (as they do in the case of 'certainty of judgment'), but rather its output.They are defined during the dialogic process.They are endogenous rather than exogenous.Learning is undoubtedly a central aspect of all deliberative processes, but here the learning aspect is absolutely dominant: deliberation attempts to take on an educational role, even before its goal of producing an exchange of opinions (Button and Mattson, 1999).
Deliberation in cases of 'uncertainty of judgment' is exposed to different types of risk.In this case manipulation, which is very improbable in the other cases, is a concrete possibility that must continually be taken into account.When the educational function of deliberation tends to predominate over the discursive aspect, participants may end up by taking positions of deference toward experts and witnesses; they become 'pupils' more than protagonists (Button and Mattson, 1999).The fact that the process is pre-structured may give it a paternalistic aura (Rostbøll, 2005): the facilitators may guide or condition the progress of the discussion, even if inadvertently.Conversely, since participants must define their own preferences during the process, a tendency toward conformism cannot be ruled out: the more uncertain participants will be induced to follow the first proposals that are formulated, those that are put forth with greater determination, or those that appear to receive group consensus.It is possible that some participants will end up by publicly taking positions that they would not have taken in other circumstances.This risk is strengthened by the pressures that tend to be created within the group to reach a unanimous consensus.These pressures can generate resolutions that formally enjoy the adhesion of all participants, but that are actually the fruit of superficial convergence and are thus fortuitous and volatile.Deliberation works if the terms of the controversy are clear and explicit (Manin, 2005), whereas in these cases there is a risk of deliberation without antagonism (too facile an operation, and fundamentally useless).The paradox of deliberation thus lies in the fact that if it takes place among people who express well defined and reflective opinions it risks ending up in sterile confrontation; and if it takes place among people who start from vague or uncertain opinions, it risks producing unstable decisions that have little credibility in the eyes of public decision-makers.
However, these deliberative processes do have one advantage: they are the fruit of an artificial design that may therefore be corrected.There is now considerable research on the institutional design of these experiments in order to overcome their drawbacks.The proposals to improve them concern all aspects of their design: extending the process and diluting it over time (Ward et al., 2003), providing the possibility for reiteration (Fung, 2003) and many other changes, including detailed ones (Carson, 2006).Research is therefore open, although it is not clear to what extent it will succeed in avoiding the more critical aspects of these experiences.

Type d: Prejudice
The last case of symmetrical deliberation is that which involves people who express well defined opinions, without however possessing a clear idea of the terms of the question.They are resolute in their options, but know little about them.Their positions are thoughtless or, we might say, are based on prejudice.
Deliberative forums of this type do not form spontaneously: people who hold prejudices are unlikely to enter into discussion with those whose prejudices are different from their own, and they are not interested in doing so.It may plausibly be supposed that deliberation among prejudiced people only takes place in the artificial contexts I presented in the previous section.Indeed, it is possible that the same experiment can, at one and the same time, combine people who have vague and confused ideas ('uncertainty of judgment') and others who have clear but unreflecting opinions ('prejudice').However, it seems likely that the very characteristics of the deliberative process would tend to discriminate between the two types of position: 'uncertainty of judgment' tends to prevail when the themes under debate are complex, unfamiliar to the general public, and have not been the subject of particularly heated controversies.Vice versa, prejudices more easily surface where hot questions are at stake for which explicit fracture lines exist in public opinion Like the previous one, this type of symmetrical deliberation essentially plays an educational role, but the direction in which it moves is, in some respects, the opposite.
There it was a question of strengthening and helping to form initially ill-defined participants' opinions; here on the contrary it a question of weakening their initially well defined positions and of sowing doubt.This is a delicate operation that can have undesired consequences.This is the deliberative context in which polarization can most easily be produced, so that participants end up by taking even more extreme positions than those they had at the outset (Sunstein, 2002).Alternatively, it may simply freeze their points of view: they may end up by only listening to arguments that are in agreement with their original positions, and ignoring stimuli that might cast doubt on those positions.In the end, fences may have been erected that reproduce the initial divergences or even strengthen them (Manin, 2005).
However, it is interesting that, in concrete experiences such as deliberative polling these negative effects do not actually come about (Luskin et al., 2002;Farrar et al., 2009).
It is not yet clear which aspects of the institutional design (providing carefully balanced information, selecting participants by lot, discussion in small randomly-drawn groups) tend to prevent this stiffening of positions (Manin, 2005).The impression remains, though, that like in the previous case the deliberative setting has a determinant influence on participants' attitudes of closure or of opening.

Models of asymmetrical deliberation
Deliberation is asymmetrical when the discussion takes place among people who, in the same context, express positions of different types: well and ill-defined, reflective and unreflective, free and constrained.Participants happen to be unequal with regard to their awareness, their resoluteness, or their constraints.Asymmetrical deliberation may take two sharply different forms depending on whether it takes place in a restricted sphere in which all participants have the possibility of expressing themselves, or in an extended sphere in which there is a clear-cut distinction between those who take the floor and those who simply listen and then express -normally by voting -their final decision.
These two types correspond to the distinction, proposed by Gary Remer (2000), between the conversation model and the oratory model.Remer observed that the notion of deliberation proposed by contemporary scholars is that of an open and informal process in which participants confront each other with their own arguments, in a rational manner and with a spirit of co-operation.Deliberation here is understood as conversation.But deliberation may also take a completely different form: a small number of orators standing in front of an audience that they are aiming to persuade.This is the type of deliberation to which the ancient thinkers referred.What Aristotle and Cicero had in mind was the use of the art of rhetoric by political leaders in a public meeting, or by lawyers in a court (Elster, 1998a;Remer, 2000;Urfalino, 2005b).The oratory (or rhetorical) model describes a deliberative situation that is completely dissimilar to that of a conversation.Instead of direct debate among participants, here we have a separation between orators and audience; the appeal to reason is overwhelmed by the appeal to passions; open and informal discussion gives way to a confrontation limited by procedural rules; co-operation is replaced by antagonism.The fundamental difference is that in conversation the participants try to persuade one another; in the oratory model the orators try to persuade a third party (the jury or the audience).In both cases the deliberative process seeks to transform preferences; but in the first case these are the preferences of the participants themselves, whereas in the second case only the listeners' preferences are concerned.
All the types of symmetrical deliberation examined above follow the conversational model, since the participants speak directly to one another with the mutual aim of persuading each other.But also an asymmetrical deliberative context can take the form of a conversation, when it comes about among a limited number of participants each of whom has the same opportunity to make his or her voice heard.Hence asymmetrical deliberation can present two different configurations: the asymmetrical conversation and the oratory model.

Type e: Asymmetrical conversation
In the asymmetrical conversation, the scenario is a contradictory one.The participants are in a condition of formal parity; they interact together within a circumscribed space; all have the equal possibility to speak; they are engaged in a discussion whose goal is to find some common ground.And yet there is a clear disparity among the participants: some of them display positions that are more resolute, more constrained and less malleable and hence they are likely to possess also better argumentative skills.
Of all deliberative processes, the asymmetrical conversation is furthest from the ideal speech situation.Control is likely to be exercised by the persons that express resolute positions over those whose convictions are more uncertain or less reflective3 .Especially if there are no external mediators or facilitators, who could try to restore balance to this disparity, the most probable outcome is a subtle process of manipulation in which the stronger subjects are capable of setting the agenda, or sanctioning arguments that are deemed inadmissible or unreasonable.
Situations of this type, in which experts, activists, politicians and ordinary citizens are involved (formally) on equal terms, are very frequent.An illuminating report comes from direct observation of some Agenda 21 forums held in the city of Antalya (Doğanay, 2004).The researcher notes that, from the moment when the debate started, there was a marked difference between experts and bureaucrats on one hand and ordinary citizens on the other: the latter 'introduced themselves and greeted the other participants at the beginning ….However, the experts or bureaucrats neither introduced themselves to the public, nor greeted the public.Because of the identification between their expert positions and personal identities, the experts constructed their authority as unquestioned and acknowledged their privileged position' (ivi, p. 739).The possibility to take the floor was likewise influenced by participants' status: 'in the meetings where the males, experts, bureaucrats and the aged participants composed the majority, the females, the 'nonprivileged' citizens and younger participants avoided speaking and hardly got a turn, even if they dared to.The moderator or other male participants mostly interrupted (…) Furthermore, discourses oriented towards women's action, which focused on acting together, had difficulties in accommodating the conflictual, abstract, argumentative styles developed by the expert, bureaucrat and male participants' (ivi, p. 740).The author concludes that 'deliberative procedures in working groups result in disregarding or rendering worthless the alternative views that challenge the established framing of the discussion.Different status perceptions of participants and the mechanisms of exclusion internal to the process generate a discursive closure that limits the democratic comprehensiveness of deliberative experiences' (ivi, p. 741).
External inclusion (i.e.people's access to the forum) can conceal an internal exclusion, which is the product of those discursive mechanisms through which people are refused the 'effective opportunity to influence the thinking of others' (Young, 2000, p. 55).
In this context, the criticism that has been aimed at deliberation, as a potentially discriminatory practice based on an imbalance between linguistic skills and, in the last analysis, on inequality, appears to be completely justified.My standpoint is that this criticism should not be addressed simply and solely 'against deliberation' (Sanders, 1997), but rather against that particular form of deliberation that is the asymmetrical conversation.
In situations of symmetrical deliberation, these risks are -as we have seen -much smaller.

Type f: Oratory model
When asymmetrical deliberation takes the oratory form, the situation changes radically.Here we are in a less restricted environment in which no-one claims the effective parity among participants.In the oratory model, the distinction between those who speak to the public (with determined and reflective initial positions) and those who listen (most probably with less defined and less reflective initial positions) is clear and generally accepted.Because the listeners are given the power to take the final decision by voting, for the orators the entire game consists of using sufficiently convincing arguments to persuade the majority of the audience to take their side.The participants/spectators have the same expectation.They take up a listening pose to understand how they should side when the oratory duel concludes.
In contemporary society, the commonest example of oratory deliberation is the public meeting.From the standpoint of modern theories of deliberation, the meeting model is not a good example.This is not only because it lends itself to demagogy, to the orators' use of insincere and strategic arguments, but also because it limits other participants to the passive role of spectator, from which only at the end, but exclusively through voting, are they released.It is significant that the participatory approaches elaborated over the last few decades systematically distance themselves from this model, aiming rather to design deliberative settings in which participants face one another directly in small groups in acceptable conditions of parity.The most visible example is the technique of the 21 st Century Town Meeting (www.americaspeaks.org)that enables thousands of people to meet in a single place to tackle a common theme.Participants are not relegated to anonymous seats in the audience, in front of the orators' stage (as in the Athenian ekklesia, or in the 18 th -century town meetings) but are arranged in small groups within the hall, helped by facilitators, and can communicate together through computers connected in a network.In other words, the meeting model has been redesigned and broken up, so that deliberation based on conversation predominates.
Modern supporters of the oratory model object, though, that in the contemporary world deliberation is still essentially based on rhetorical confrontation.The oratory model lends itself better to explaining what happens in real decision-making: it is less demanding and more realistic (Yack, 2006) in a mass democracy (Chambers, 2009).
Furthermore, it presents a fundamental advantage over the conversation model: that of placing antagonism at the centre of deliberation.Bernard Manin (2005) observes that processes of the discursive and dialogic type (i.e.conversations) may produce disappointing results, since interaction among the participants, may produce conformism or strengthen initial convictions.For deliberation to take place it is not enough to open a discussion among differing opinions.It is essential that participants be faced with a clearcut contraposition, through which all the pros and cons are made explicit.In this sense, the oratory model offers better guarantees, because although it does not allow the participants to interact together, it does give them a clear picture of the terms of the controversy.
Antagonism, even without dialogue, is preferable to discussion without antagonism.
In reality, the two models can perfectly well be integrated together (as indeed is suggested by Manin).There is nothing to prevent the deliberative process from being subdivided into two separate phases: an antagonistic phase in which stakeholders or experts express their contrasting positions in front of the meeting, and a dialogic phase in which the participants meet for discussion among themselves.In this way '[the] participants make their choices based on the arguments they have listened to, but also on the arguments they have exchanged.The decision results from confronting reasons, and not only through aggregating preferences.But the meeting will also have weighed the pros and the cons of its decision, which a simple discussion would not have been able to ensure' (Manin, 2005, p. 191).We may add that this mixed model reflects very closely the millenary structure of the criminal trial, and nor is it particularly dissimilar from recent experiments, such as citizens' juries and deliberative polls, that separate the phase during which arguments are expounded from that in which the participants reason among themselves.
reflects the increased interaction that is becoming necessary to make any type of public decision.To recognize the different nature of deliberative processes may thus be a significant step in designing specific institutional settings capable, at least to some extent, of governing them and holding in check their possible negative effects.
correspond to the four types of position we examined above: a) suspension of judgment; b) certainty of judgment; c) uncertainty of judgment; d) prejudice.For each of these four types of symmetrical deliberation I will examine: the characteristics of the participants' positions; which empirical cases tend to correspond to the model; how the deliberation takes place; what risks it runs; what mechanisms can limit those risks.Type a: Suspension of judgmentThe first type of symmetrical deliberation comes about among people with a high degree of awareness of the problem and of the controversies surrounding it, but who despite this (or just because of this) prefer to suspend their judgment.The positions they take are ill-defined, open and flexible.We may define such persons as skeptics or agnostics.They have a strong orientation towards agreement and co-operation.They are open to dialogue.They are like the Socratic sage who knows that he does not know.