Guilty Plea Decisions: Moving Beyond the Autonomy Myth

When a defendant pleads guilty to a criminal charge against them their conviction may be justified on the basis of autonomy rather than accuracy. In this context, autonomy can make the difference between a legitimate conviction and the breach of fundamental rights. However, autonomy in this context is not clearly defined. This article argues, based on philosophical conceptions of autonomy and empirical realities, that true autonomy is an ideal rather than a practical reality. It considers the level of autonomy necessary to legitimise a criminal conviction via plea, and suggests that current conceptions of autonomy are inadequate since they rely on a formalistic autonomy ‘myth’, presuming autonomy in the absence of threats. An analysis drawing on original empirical data from two studies demonstrates how autonomy may be being depleted to unacceptable levels in the current system. The article concludes by presenting reform proposals.


INTRODUCTION
When a defendant in the criminal justice system decides to plead guilty this constitutes a waiver of their right to a full trial under Article 6 of the European Convention on Human Rights (ECHR).The prosecution is no longer required to prove that they are guilty beyond reasonable doubt, and their status automatically changes to that of a convicted person.It is not required that a guilty plea be shown to be an accurate admission of guilt, or even reliable evidence of guilt.As Jeremy Horder states in Ashworth's Principles of Criminal Law: 'A guilty plea may simply be accepted more or less at face value, as proof beyond a reasonable doubt of the allegation in question … A guilty plea is not automatically subject to vigorous testing for its veracity, in the way that a plea of not guilty is tested.' 1 Because a guilty plea does not need to be tested in the way that a plea of not guilty does, guilty pleas have important benefits for the criminal justice system.When a defendant pleads guilty, it can reduce the impact of crime upon victims, it saves victims and witnesses from having to testify and it saves public time and money on investigations and trials. 2In fact, if all defendants in England and Wales wanted to go to trial rather than plead guilty, the criminal justice system would be in danger of collapse due to the time and expense involved in trials. 3However, this lack of testing for veracity also means that convictions via guilty plea cannot necessarily be justified in the same way as convictions via trial, which are typically justified on the basis of being both accurate and obtained in a fair and rights-compliant way. 4 Having sufficient justification for convicting a defendant is vital from a rights-based perspective, since convictions involve significant curtailment of fundamental rights guaranteed by the ECHR, including the right to liberty and the right to respect for private life.Such rights cannot be curtailed purely to attain benefits for the criminal justice system.
Reliance on guilty pleas to achieve criminal convictions despite the absence of meaningful enquiry into their veracity can potentially be justified in one of two ways: first, if only guilty people are likely to plead guilty and thus the guilty plea can be accepted as a reliable indication of guilt absent additional enquiry in individual cases, or, second, if convictions arising from guilty plea are justified on a basis other than the guilt of the defendant, for example on the basis of defendant autonomy.Guilty pleas have traditionally been viewed as reliable indications of guilt, in part because they are seen as admissions made against one's own interest.However, the presence of incentives to plead guilty mean that a guilty plea may not always, or even typically, be an admission made against one's own interest.Research suggests that both guilty and innocent defendants do plead guilty, and legal systems often either explicitly or implicitly acknowledge that this is the case. 5In England and Wales, data provides support for the contention that people who are factually innocent, or at least who may not be legally guilty, are pleading guilty.Of the 128 cases referred to the Court of Appeal by the Criminal Cases Review Commission as potential miscarriages of justice (for review of conviction only or of conviction and sentence) from 2012-18, approximately 50 cases involved defendants who initially 2 Sentencing Council, Reduction in Sentence for a Guilty Plea Definitive Guideline (2017), 4 at https://www.sentencingcouncil.org.uk/overarching-guides/crown-court/item/reduction-in-sentence-for-a-guilty-plea-first-hearing-on-or-after-1-june-2017/#B.%20Key%20principles (last accessed 28 June 2021).3 R v David Caley and others [2012] EWCA Crim 2821 at [6]. 4 I. Dennis, The Law of Evidence (London: Sweet and Maxwell, 6 th ed, 2017) ch 2. Use of guilty pleas can be seen as part of a trend away from a 'rights-bearing' defendant in the criminal trial, see R. Gimson, 'The mutable defendant: from penitent to rights-bearing and beyond' (2019) 40 Legal Studies 113.  5 In the United States this is explicitly acknowledged by what is known as an Alford plea through which an individual can plead guilty while maintaining a protestation of innocence.See C. Shipley, 'The Alford plea: A necessary but unpredictable tool for the criminal defendant' (1986) 72 Iowa Law Review 1063.Note that this is not the case in all jurisdictions; for example Canada require an accused to admit all elements of an offence to plead guilty, although this is not thought to stop the innocent pleading guilty, see J. Brockman, 'An offer you can't refuse: Pleading guilty when innocent' (2010) 56 Criminal Law Quarterly 2010.
pleaded guilty. 6Anecdotal evidence from individual cases can also provide support for the contention that innocent defendants do plead guilty in England and Wales.For example, consider the case of Michael Holliday.Holliday was convicted of a robbery on the basis of a guilty plea which he claimed was entered on the advice of his lawyer. 7On appeal, clear evidence was presented that someone else committed the offence and evidence showed that admissions made by Mr Holliday were not reliable.The court described evidence of the unsafety of the conviction as 'very strong indeed.' 8 The reduced importance of factual guilt or innocence in the guilty plea context can also be seen when looking at the Code for Crown Prosecutors guidance on accepting guilty pleas to fewer or lesser charge(s) than those a defendant is accused of.Specifically, the code requires prosecutors to consider the needs of the victim, the sentencing options available, and the seriousness of the offending, but there is no requirement to consider the accuracy of the resulting conviction. 9In fact, it could even be considered contrary to the spirit of the adversarial model to expect the prosecution to shoulder any such responsibility or burden, since the defence lawyer rather than the prosecutor typically has the responsibility to defend a client, to make the case for his or her innocence, and to advise against pleading guilty when innocent. 10herefore, it is not appropriate to rely on the assumption that only guilty defendants plead guilty when seeking to justify conviction via guilty plea. 11n a system in which guilty pleas are an accepted and important part of the criminal justice system, it is consequently necessary to consider other ways in which convictions via guilty plea might be justified.In a recent article, Richard Nobles and David Schiff conclude that autonomy, rather than factual accuracy, justifies the conviction of defendants via guilty plea. 12They state that: The presentation of procedures of the modern criminal justice system as giving priority to the truth of convictions or avoidance of wrongful conviction misrepresents the system, whose legitimation can be better represented by reference to the concepts of autonomy and the autonomous exercise of rights. 13his conclusion is consistent with regulations surrounding guilty pleas.For example, convictions obtained via guilty plea are harder to appeal than those obtained via trial despite not being proved to the same level. 14This treatment can be justified on the basis that convictions via guilty plea are legitimised by defendant autonomy in addition to, or even in place of, factual accuracy.When a guilty plea decision is entered autonomously, it can be seen as disrespectful to a defendant to not accept that decision and to force them into further proceedings. 15The importance of autonomy in legitimising plea decisions is also consistent with jurisprudence in the ECHR context where there is no requirement for a plea to be an accurate indication of guilt, but a plea must be 'established in an unequivocal manner, and be given in full knowledge of the facts, that is to say on the basis of informed consent … and without constraint …' 16 The requirement of informed consent without constraint is closely tied to the concept of autonomy, since consent is typically viewed as a form of autonomous choice, aimed at authorisation. 17 In this context, defendant autonomy makes the difference between a conviction being legitimate or illegitimate, and consistent or inconsistent with fundamental rights.However, despite its vital role in the guilty plea system, autonomy is not clearly defined in this context.Developing a clear definition of what exactly is required for a plea decision to be autonomous is important because of what is at stake and dependent on autonomy and also because of potential threats to autonomy in a system where the state both sets the parameters of plea decisions and benefits the most when defendants plead guilty.In this system, a robust and empirically appropriate definition of what is required for a decision to be 'autonomous' is necessary to protect defendants.However, such a definition is not provided in the current system.This article draws on philosophical conceptions of autonomy and empirical realities to (i) recognise autonomy as an ideal rather than a practical reality, (ii) propose a level of autonomy necessary to legitimise a criminal conviction via guilty plea, and (iii) evaluate current practice in light of that proposal.
The first part of the article considers and critiques varying philosophical conceptions of autonomy, and identifies the relational approach to autonomy as the approach that most fully embraces the normative underpinnings of autonomy and related empirical realities in allowing individuals self-governance.The second part of the article examines the conception of autonomy in the guilty plea process.It develops a standard for the minimum level of autonomy necessary to legitimise a conviction via guilty plea and demonstrates that the current system relies on an autonomy 'myth' rather than a meaningful enquiry into autonomy.
14 Although note that there is also more that can go wrong at a trial and give rise to a right to appeal, for example misdirections may be given by judges or the rules of evidence may not be adhered to.15 For relevant accounts of respect see S.L. Darwall, 'Two kinds of respect ' (1977)  The final parts of this article analyse original empirical data to provide insight into the autonomy of those making plea decisions in the current system.This analysis suggests that the level of autonomy in plea decisions is insufficient to legitimise convictions via guilty plea, and proposals for reform are developed.

CONCEPTIONS OF AUTONOMY
Autonomy is typically understood as meaning self-determination or selfgovernance.That is, the freedom to pursue one's conception of what is good or desirable, as long as it does not impinge on the freedom of others. 18So a choice that does not reflect a chooser's goals or values is not an autonomous choice where these goals or values represent the core of the person. 19Autonomy is thus the opposite of oppression or coercion,where individuals are guided by external factors and not their authentic beliefs or desires.However, varying conceptions of the precise nature of autonomy in the context of decision-making can be identified.Three central conceptions are outlined below.These conceptions can form a scale of less to more cohesive requirements underlying autonomy, and thus informed consent.

Autonomy as the ability to make a choice
According to this conception, autonomy can be viewed as the ability to make a choice, meaning the capacity to make a choice (an internal condition that is made up of competencies such as understanding, appreciation, and ability to reason that allow a person to make a decision that reflects their goals) 20 and the absence of coercion that entirely restricts choice (an external condition, often required to be an external threat).Put simply, a person lacks autonomy if they have no capacity to make a choice or if they have the capacity to make a choice but not making a decision is not a viable option (for example because making a decision a certain way would have very negative consequences).This type of approach can be seen in theoretical bases for the defence of duress in the English criminal law.A defendant who has committed a criminal act (other than murder) will be found not criminally liable for their actions where they committed the crime because of threats of death or grievous bodily harm, and a reasonable person would have acted as they did. 21One justification that has been offered for this defence is that the defendant lacked choice when deciding whether to commit the crime.The defendant did not have a fair opportunity to do otherwise. 22 choice because they constrain a person's choice in a way that other causes do not. 23iewing autonomy in this way means that decisions are made autonomously provided that they are made by someone with the capacity to make them, with a meaningful choice to make (usually meaning the absence of external threats).However, such a conception of autonomy is limited and inconsistent with philosophical conceptions of personal autonomy.Perhaps most importantly, this conception over-estimates the ability of people to act freely in pursuit of their goals by presuming a person can act autonomously provided that there is a choice that they can make.In reality, the fact that a person can make a choice does not mean that they are able to make a choice that is free and in accordance with their own perceptions of what is good.The limitations of ability to choose models of autonomy have been recognised by Kathryn Hollingsworth in the context of children's rights.Hollingsworth classifies the capacity for choice as agency, which can be distinguished from full autonomy.She notes that in the context of a rights-based punishment system,requiring only agency of decisionmakers is problematic, in part because of an assumption that all individuals are equally able to act freely in making choices. 24Literature in the context of sexual offences has also made parallel arguments, calling for meaningful enquiry into the related concept of consent, beyond an enquiry into whether a complainant said 'yes' or 'no.' 25 Martha Fineman's vulnerability theory can provide additional insight into the reasons that autonomy may be constrained despite the ability to make a choice.The theory describes the inherent vulnerability of people to depletions of autonomy in both overt and subtle ways.According to vulnerability theory, all people are vulnerable due to their embodiment (that is, their existence as visible and embodied human beings, and how these material bodies interact with the world) and their embeddedness (that is, their place within relationships and institutions). 26In this context, individuals do not make decisions in isolation but instead decisions are tempered and curtailed by an individual's position within social, legal, and economic institutions and relationships.For example, people cannot gain employment in a career they feel is worthwhile without access to necessary education.People are therefore inherently dependent on various relationships and institutions (to which people have varying degrees of access) in order to facilitate freedom in decision-making.This view of the human condition sits in contrast to the conception of individuals as autonomous A more expansive version of autonomy, based on what have been termed coherentist approaches to autonomy, holds that a person makes an autonomous decision where they make a decision in accordance with their own point of view. 28Again, this will involve both internal components (relating to the person themselves) and external components (relating to their external environment).Put simply, this account hold that a decision cannot be described as autonomous where a person themselves repudiates or disassociates themselves from what they are doing. 29If this happens then the power of the motives that the agent has for making their decision is independent of their authority.So, for example, a drug addict would not be acting autonomously if she failed to govern an irresistible urge to take drugs, but would be autonomous if she had no objection to her addiction and its effects. 30One proponent of this type of approach is Joseph Raz, who argues that autonomous people identify with their choices, and that people who are alienated from their motives and the forces that drive them are not truly autonomous. 31There is therefore an important distinction in coherentist accounts between motives or first-order desires (what an agent wants to do now) and second-order desires (the desires that an agent reflectively has about what they want or what is good).For a decision to be autonomous, it is necessary (but not sufficient) that there be consistency between the decision and second-order desires.So, if an individual strongly opposes taking responsibility for an offence that they have not committed, an incentive leaving them to choose between pleading guilty when innocent and not being able to provide from their family could be said to undermine autonomy, despite the presence of choice.This concept of autonomy is far more wide-ranging than autonomy as the ability to make a choice, and recognises that a variety of factors can impede self-governance without denying a person the ability to make any choice.It is therefore far more consistent with the empirical realities of choice, particularly given the inherent vulnerabilities of individual decision-makers noted above.However, more modern accounts of autonomy recognise additional components necessary to facilitate true self-governance and therefore the full requirements of autonomy.
Autonomy as the ability to conform an act to a mental state that represents one's own point of view, and the psychological and social conditions to support the exercise of this ability in practice Relational accounts of autonomy have added to the coherentist approaches of autonomy discussed above.Such approaches suggest that autonomy can be dependent on the environment that a person is in and certain attitudes that a person has toward themselves.First, relational accounts have identified the importance of self-trust, self-respect, and self-esteem in autonomous decisionmaking.These traits are to some extent internal, but in reality, are facilitated (or depleted) by the external environment that a person is in. 32These traits influence the capacity of an individual to act autonomously, since they facilitate a sense that there is a point to what a person is doing, and allow an individual to assess and reassess their motivations and to make decisions in alignment with them.Without self-trust, self-respect, and self-esteem a person's sense of the worth of their aspirations is reduced and autonomy is hampered. 33This effect has been empirically demonstrated by psychological research, for example work on purpose and 'derailment'. 34On this basis, a person's social environment, in addition to the specific aspects of the decision that a person is making, is key to maximising autonomous decision-making.Additionally, relational accounts have noted the importance of a person's place in society and the effect that this is likely to have on the options available to them and their ability to identify and assess those options.For example, the capacity for women to exercise their autonomy may be hampered or even eliminated by systems of gender inequality. 35Relational autonomy in the legal context has been explored in the work of Jennifer Nedelsky, who describes autonomy as a capacity made possible by constructive relationships, and the law as a tool for shaping relationships that can foster or undermine autonomy. 36hese relational conceptions of autonomy most closely mirror the empirical realities of autonomy where, for reasons described above, individual choice is inextricably linked to social context.

DEFINING AUTONOMY IN THE GUILTY PLEA PROCESS The normative importance of autonomy in the guilty plea process
As detailed in the introduction, autonomy can provide a potential justification for convicting defendants in the criminal justice system on the basis of a guilty plea.Essentially, while a conviction at trial is a conviction based on evidence against a defendant, a conviction via plea can be seen as a conviction based on the autonomous choice of the defendant.Since the defendant is the person affected by the conviction, it is arguably right that their choice to be convicted without a full trial be respected. 37To force a trial on a defendant who does not want one can be seen as disrespectful and paternalistic in an adversarial system, particularly where the defendant clearly claims they are guilty.However, it should be noted that even autonomous guilty pleas can be harmful when entered by innocent defendants, particularly where true offenders avoid justice as a result.
In analysing the role of autonomy in justifying convictions, Nobles and Schiff review the work of Ronald Dworkin in order to show that even work routinely cited in support of the need to prefer acquitting the guilty over convicting the innocent offers no basis to resist incentivised autonomous pleas. 38They suggest that autonomy may be an adequate basis for a conviction since when an individual autonomously decides to forego their right to a full trial, they have received the level of protection (particularly against wrongful conviction) offered by a full trial, but have chosen to surrender it.This protection means there is no problem with equal respect and concern since individuals are treated equally. 39hus, autonomy may have the potential to justify convictions even where these convictions are not justified by traditional legitimations of punishment which include a requirement that the conviction be accurate (ie only guilty defendants are convicted). 40n this context, autonomy is important both in transforming a violation of fundamental rights (for example through the restriction of liberty or invasion of private and family life of a person who has not been convicted legitimately) into appropriate state action, and in ensuring equal protection for criminal defendants.For two primary reasons, the concept can only carry this weight if it is interpreted in an empirically reasonable way that reflects true self-governance.
First, because depriving a person of their fundamental rights carries huge normative and practical significance and requires the strongest form of autonomy to be justified.If a person pleads guilty but feels dissociated from the reasons for doing so, they have given up their rights for reasons they do not truly associate with.Of course, individuals are required to make decisions all the time that are impacted by their vulnerability and by restrictions on their choices.The important distinction in the guilty plea context is that individuals are waiving a fundamental human right and being subjected to punishment and stigma by the state on the basis of autonomy, thus it is essential that decisions are truly autonomous.In no other context are individuals encouraged to give up fundamental rights as a result of pressures directly created by government action.It may be argued that in this context defendants may have committed a crime and in doing so have surrendered their claim to autonomy, however this line of reasoning is inconsistent with the fundamental principle that up until the point of conviction every defendant should be presumed innocent.
Second, only if a realistic conception of autonomy is used can equal protection for defendants be ensured.Conditions with the potential to deplete autonomy and pressures to make decisions a certain way (even when not external threats) have the potential to influence groups differently along various axes of power and prejudice.For example, the threat of even a short period in custody may disproportionately influence single parents who have no one else to care for their children.Existing research already shows systematic differences in plea decisions among different ethnic groups, with minority groups choosing to plead guilty less often. 41This effect has been explained as resulting from a lack of confidence in the criminal justice system amongst minority groups. 42Here, there is a clear risk that defendants in such groups may be making decisions based on fear or anger in an environment that is not conducive to self-trust, self-respect, and self-esteem and therefore not conducive to autonomy.
Only the use of a conception of autonomy that recognises the full range of potential challenges to self-governance can truly protect defendants and their right to a fair trial.Using such a full conception of autonomy is important because of the practical reality of the guilty plea context in which the interaction of inherent vulnerabilities of criminal defendants and incentives to plead guilty have the potential to challenge autonomy in important ways that do not involve external threats.
First, defendants in the criminal justice system are inherently vulnerable as a result of the imbalance of power between themselves and the state. 43Even before conviction of any offence,the state is able to curtail their rights,including their right to liberty and their right to respect for private life.This curtailment may simply include requiring the defendant to attend court at a specified date or time, or may include arduous bail conditions.In the most extreme cases this can include holding a defendant in pre-trial detention pending trial.As the Crown Prosecution Service notes, 'from the viewpoint of the defendant, bail decisions made by a court can result in the deprivation or restriction of liberty for a substantial period of time.' 44 This deprivation, while often necessary for the interests of society, can mean that defendants rights are restricted before they have even been found guilty.In the last quarter of 2018,the average waiting time for trial in the Crown Court was 14.8 weeks,rising to 29.2 weeks for defendants pleading not guilty (although waits for those in custody were roughly half as long as waits for those released on bail). 45Many defendants may have family or employment circumstances that make such restrictions on freedom particularly harmful.For example, a report examining female offenders in 2012 noted that between 25 per cent and 31 per cent of female offenders were believed to have one or more child dependent(s). 46Defendants also frequently have limited resources and have the threat of conviction at trial hanging over them.These factors highlight potential threats to autonomy in criminal defendants, both in terms of potential direct pressures and in terms of potential undermining of self-trust, self-respect, and self-esteem.
Second, these defendants face incentives to plead guilty.Whilst incentives are typically viewed in a positive light and provide positive opportunities for defendants, they have the potential to create both explicit and implicit pressure to surrender fundamental rights and can, as noted above, be viewed as creating penalties for refusing to surrender fundamental rights.In the academic literature on guilty pleas, there is a lack of consensus on whether shorter sentences when pleading guilty should properly be categorised as plea discounts or trial penalties. 47Where defendants know the sentence they will receive, or will be likely to receive, if pleading guilty, this may become the norm from which they evaluate potential outcomes at trial.This outcome is particularly likely in systems such as England and Wales where the majority of defendants plead guilty, and thus obtain the reduced sentence.As a result, defendants may view trial as a 'loss' compared to the plea option, rather than the plea option as a 'gain' when compared with trial.Sentence differences that the system views as incentives to plead may then be viewed by defendants as penalties for going to trial, potentially generating additional pressure to plead.The importance of incentives to plead guilty and their potential to influence defendant autonomy was highlighted by Mike McConville and Luke Marsh, who have described a system of 'state-induced guilty pleas.' 48he criminal justice system in England and Wales offers direct incentives to defendants to encourage them to plead guilty, and to do so 'as early as possible.' 49These incentives are stated to be 'directed only at defendants wishing to enter a guilty plea' and are not intended to 'put pressure on defendants to plead guilty.' 50Specifically, Sentencing Guidelines provide for an up to one-third reduction when a defendant pleads guilty at the earliest opportunity, reducing to an up to one-tenth reduction when a defendant pleads guilty at the beginning of a full trial. 51Importantly, this reduction can change the type of sentence imposed on a defendant, from a custodial sentence to a community sentence or from a community sentence to a fine.The Crown Court Sentencing Survey indicated that, in 2014, 89 per cent of Crown Court defendants who pleaded guilty at the earliest opportunity were awarded a sentence reduction of 33 per cent or more. 52Empirical research examining this survey data has suggested general judicial compliance with suggested sentence reductions, with some variation among offences. 53However, while the guidelines have been developed to avoid putting pressure on defendants to plead guilty, the guidelines have not drawn on empirical work to identify how and when such pressures may arise and to ensure that they are not created.
It is also possible for defendants to obtain reductions in the charge(s) against them when they plead guilty.These charge reductions occur where a prosecutor is willing to drop a more serious charge against a defendant if they agree to plead guilty to a less serious charge.These reductions are regulated by the prosecutorial guidelines contained in the Code for Crown Prosecutors. 54eductions have the potential to change the penalties that could be faced by the defendant. 55Importantly, in England and Wales defendants can gain a relatively good idea of their likely sentence if they plead guilty because of the relative predictability of sentence discounts, and the ability to gain an indication of the likely maximum sentence that will be imposed if a guilty plea is entered in Crown Court cases, through a Goodyear hearing. 56The combination of predictable sentence discounts and the availability of Goodyear hearings is designed to provide a defendant with clarity and so enhance certainty and compliance with the rule of law.However, confidence in a significantly lesser sentence if pleading than if convicted at full trial has the potential to turn the plea decision into a direct comparison of sentence magnitudes, rather than a qualitative decision influenced by guilt and values.Defendants can also be influenced by informal incentives to plead guilty.Such incentives include the ability to have a case processed faster 58 and more cheaply (primarily, costs of legal representation are significantly higher when pleading not guilty) 59 and even the ability to obtain immediate release from custody if a guilty plea is entered. 60In the case of fines, admitting guilt through paying early can also result in significant financial reductions.
Thus, the need for the concept of autonomy to be robust enough to justify the deprivation of fundamental rights and to ensure equal protection, combined with the clear practical threats to autonomy in the plea context, necessitate the use of the most empirically realistic conception of autonomy (what it means to have self-governance) that recognises the various ways in which autonomy can be undermined.Therefore, when considering the legitimacy of guilty plea decisions, it is necessary to engage with autonomy in the relational sense (autonomy as the ability to conform an act to a mental state that represents one's own point of view, and the psychological and social conditions to support the exercise of this ability in practice).
The modern criminal justice system is clearly not designed to require fully autonomous action in this sense in order for criminal liability to be imposed.In fact, only an extreme deprivation of autonomy can excuse a person from liability for a criminal act.We impose criminal liability where an individual has engaged in criminal behaviour as a result of economic need,and the formulation and operation of current laws may even disadvantage groups that are pushed towards crime by economic and social circumstances. 61In fact, criminal law itself is largely reliant on what is known as 'folk psychology,' according to which mental states, such as desires, beliefs, intentions, and reasons, genuinely cause behaviour, and individuals as agents are responsible for that behaviour in the absence of rationality deficits. 62owever, it is not inconsistent to argue for a more meaningful concept of autonomy to be used in the guilty plea process while maintaining current conceptions of responsibility in the criminal law.In the guilty plea process, but not in more general decision-making, pressures with the potential to undermine autonomy are being introduced and imposed directly on an individual by state action.The party who can benefit most clearly from the decision of the defendant to plead guilty (the state) is setting many parameters of the plea decision, and controlling the conditions in which a defendant is making the plea decision.By setting parameters of the decision, the state is able to exert control over

Developing a practically useful conception of autonomy
There remain significant challenges to recognising such a full concept of autonomy in the guilty plea process.True autonomy in this sense is likely to be impossible to fully attain, both generally and in the guilty plea context.To some extent, the concept is a metaphysical ideal rather than a practically useful concept.As Gerald Dworkin stated in the context of defining autonomy, it is not feasible to 'specify necessary and sufficient conditions without draining the concept of the very complexity that enables it to perform its theoretical role.' 63 There are many uncontrollable influences in people's lives and aspects of individual identity can be shaped and even controlled by their environment.This reality has led Martha Fineman, the founder of vulnerability theory, to conclude that true autonomy is essentially a 'myth'. 64What is needed to apply autonomy in practice is therefore a post-metaphysical, pragmatist approach that is applicable to the specific context and its normative requirements. 65In this context, autonomy is not truly the dichotomous concept that it is portrayed as by the current system.Rather autonomy is a scale, from non-autonomous action to fully autonomous action (an ideal rather than practical reality).It is therefore important to determine the level of autonomy that is necessary to legitimise a conviction via guilty plea.It is also important to define autonomy in a sense that can be applied in a predictable way by courts, to avoid judges having to make uncertain ethical decisions.So, what does a sufficiently autonomous plea decision look like?A sensible focus when assessing autonomy in a plea decision is the 'core' of the plea decision itself, and the ability of the defendant to exercise their second-order values applicable to it.This focus can ensure defendants can act in accordance with their own sense of what they want or what is good or fair (their second-order desires) without involving judges in an ethical enquiry into the sources of those values themselves.At its core, the plea decision is a decision whether to self-incriminate or to contest guilt at trial.The second-order values relevant to this decision are likely to vary person-to-person, and are likely to vary based on whether a person is guilty or innocent. 66For example, a person who is guilty might recognise that it is good to admit to something you have done wrong but also that it is good to avoid criminal conviction; while a person who is innocent would likely strongly feel that it is not good to take responsibility and blame for something that you are not responsible for.A decision can still be autonomous when influenced by incentives to plead, provided those incentives are having an influence within a framework of application of those relevant second order values.For example, a modest sentence incentive might tip a guilty defendant towards the 'good' of admitting wrongdoing rather than the 'good' of avoiding conviction.In this way, incentives are viewed and considered in light of the second-order values applicable to the 'core' plea decision.However, a decision should not be seen as autonomous where incentives to make the decision a certain way draw the defendant away from the second-order values relevant to the core decision, and come to dictate the decision themselves through invoking completely new and competing second-order values.For example, a threat of custody at trial that can be avoided by pleading guilty might draw an innocent defendant away from second-order values relevant to the 'core' decision such as not wanting to take responsibility for a crime not committed and mean a decision is determined by distinct second-order values, for example fear of being taken away from children or losing employment if convicted at trial.This approach has the advantage of capturing autonomy in a practical way that is also relatively consistent with another key feature of the criminal lawthe presumption of innocence.The formulation respects the presumption of innocence since if a person has strong second-order values relating to innocence, they should not (under this conception) be drawn away from those values by compelling competing incentives or conditions restricting their ability to decide in accordance with them.In this way, the formulation may also serve to minimise guilty pleas from innocent defendants and therefore minimise the extent to which true offenders can escape justice as the result of others pleading guilty.
If a defendant decides to plead guilty as a result of second order values not relevant to the core plea decision, the decision should not be viewed as sufficiently autonomous to justify conviction.Therefore, in convicting a defendant on the basis of a plea decision, the state must ensure that the defendant is able to make a decision that accords with these second order values.Ensuring this ability is likely to involve monitoring the incentives that a defendant may face to plead guilty and also promoting the self-trust, self-respect, and self-esteem necessary to facilitate decisions in accordance with relevant second-order values.

Current conceptions of autonomy in the guilty plea process
Current law and procedure do not protect defendant autonomy in the way discussed above.Instead, the conceptions of autonomy relied on in the literature and case-law presume the presence of autonomy in the absence of external threats.In considering whether guilty pleas are in fact voluntary, Nobles and Schiff state that 'normally one would only describe something as coerced if the harm that influences a person to make a particular choice is impermissible.' 67 On this basis, where evidence justifies a prosecution it is not impermissible for a defendant to have to make such a choice.The fact that this choice is easy does not make it coercive.
This formulation is generally consistent with recent case-law on guilty plea decisions in England and Wales and the European Court of Human Rights (ECtHR).In the case of McKinnon v United States, the House of Lords considered the case of a computer hacker (the appellant) who was facing extradition to the Unites States.The appellant was offered a plea deal whereby he could plead guilty and receive a sentence of three or four years of imprisonment, which could likely be served in the United Kingdom after six to 12 months in the United States.If he did not plead guilty he was told he could expect to receive a sentence of at least eight to 10 years of imprisonment, all of which would be served in the United States.The House of Lords found that the discrepancy between the sentence when pleading guilty and sentence if convicted at trial did not constitute an abuse of process.They noted that only in an extreme case would they regard encouragement to plead guilty as an abuse of process.To constitute an abuse of process, the threat would need to go significantly beyond the defendant's 'just deserts' on conviction, or to involve threats of unlawful action. 68Thus, the House of Lords seem to be adopting an 'autonomy is choice' approach, holding that only threats (and only relatively extreme threats) have the potential to undermine the autonomy of the defendant's decision-making process.
Similarly, in the case of Natsvlishvili v Georgia, in the context of company law offences, the ECtHR held that a defendant who would face a likely lengthy period in poor custodial conditions if he went to trial but could avoid custody if he pleaded guilty was not constrained in his plea decision-making. 69In reaching this decision, the ECtHR noted that the applicant's acceptance of the plea was 'undoubtedly conscious and voluntary,' and that the decision could not have been said to have resulted from duress or false promises. 70Again, the court seems to suggest that duress or false promises are necessary to undermine the autonomy of the defendant and thus their waiver of a right to a fair trial.
The lack of meaningful engagement with the concept of autonomy in the guilty plea process can be contrasted with the more meaningful and thorough enquiry into the concept (primarily the related concept of consent) in other areas of the law such as sexual offences 71 and medical consent. 72There is also a precedent for use of a more meaningful conception of a similar concept (voluntariness) that more closely mirrors the suggested conception in scrutinising guilty plea decisions.In the Canadian legal system, a plea is considered 'voluntary' if the accused makes a conscious and volitional decision to plead guilty for reasons he or she sees as appropriate. 73elying on the superficial conception of autonomy currently employed is not sufficient given the normative weight attached to autonomy in the guilty plea process.This reliance reduces the requirement for autonomy to a relatively formalistic procedure, relying on an autonomy "myth," and will lead to a finding of autonomy in all but the most severe cases.This approach is detached from more modern and empirically-valid conceptions of autonomy and the philosophical realities underlying autonomy as a psycho-legal concept, and has the potential to allow plea decisions that are made directly as a result of state action depleting defendant autonomy.

GUILTY PLEA AUTONOMY IN PRACTICE
Recognising the mismatch between the way autonomy is currently viewed in the guilty plea context (as choice in the absence of external threats) and the more empirically and philosophically robust requirements that have been suggested (ensuring defendants can make decisions driven by second-order values relevant to the core plea decision) provides a new context through which to evaluate current guilty plea practice and procedure.Empirical investigation of the current process can provide insight into the extent to which autonomy may be being depleted unacceptably in the current system.
This section presents the results of two empirical studies examining defendant plea decisions in practice.This analysis highlights ways in which autonomy is depleted in defendants making plea decisions.

Participants and methodology
1 examined practitioner perspectives on when and why clients decide to plead guilty. 74The data in Study 1 was gathered in September 2017 from an online survey completed by 90 legal professionals practicing criminal law in England and Wales. 75Sixty-three per cent were solicitors (including nine solicitor advocates), 29 per cent were barristers, and the remainder were paralegals /caseworkers/trainee solicitors.Seventy-four per cent of participants worked only in defence, 25 per cent in both defence and prosecution, and one per cent in prosecution only.Participants were asked a range of questions relating to guilty pleas.For the purposes of this analysis, responses to the specific question 73 R v Smoke [2017] SKQB 345 (CanLII) 74 Note that some data from this study (responses to several specific questions) has already been published, see n 6 above.The data analysed here is discrete from that data and has not previously been published.75 This represented approximately 12 per cent of the professionals identified via an internet search who were sent the survey link.
'In your experience, what is important to a client when deciding whether to plead guilty?' will be considered.All responses were anonymous.There are some limitations of this data that should be noted.The fact that the majority of participants work only in defence skews the sample, however this was not considered problematic since lawyers working in defence are likely to have the most in-depth knowledge of reasons for pleading due to the lawyerclient relationship.The fact that the response rate was around 12 per cent also reduces the generalisability of the data.For this reason, the viewpoints may not be representative (for example because those who have problems with guilty plea procedure may have been more likely to respond).However, the sample size was large enough that key trends rather than anecdotes could be identified.Finally, the lack of defendant participants in the sample limits the conclusions that can be drawn about defendant experience, and future work should target these perspectives specifically.
Study 2 gathered more detailed qualitative data examining how various aspects of the criminal process are perceived by legal practitioners.The data in study 2 was gathered from November 2018 to May 2019 from in-person interviews with 36 legal professionals, all based in Wales (some of whom practiced exclusively in Wales and some of whom practiced in both Wales and England). 76This included 20 solicitors (practicing criminal defence) (DS1-DS20) across 16 firms, and 16 barristers (practicing both criminal defence and prosecution) (BS1-BS16) across five chambers.Participants were asked a range of questions relating to how clients experienced the criminal process and whether -and how -they interacted with their legal advisers. 77This analysis focuses specifically on discussions relating to plea decisions made by clients.Participants were asked two questions on this topic: 'What are your views on the sentence discount principle?' 78 and; 'How do you think your clients experience the process of entering a guilty plea?' The use of general questions ensured free responding in participants and avoided leading participants to focus on areas thought to be important by the interviewers rather than themselves.Interviews were anonymised to protect the identity of those taking part.There are three limitations that should be noted.First, as with Study 1, participants worked principally in defence, although the barristers interviewed also engaged in prosecution work, and defendant perspectives were not captured.Second, the study relies on interview data which will likely reflect personal perspectives as well as practical realities.Third, generalisations cannot be made because of the relatively small sample size so, as with Study 1, the sample may not be representative across the England and Wales jurisdiction.However, combining the rich qualitative interview data with the detailed survey results allows us to note key trends.Data from both studies was analysed using a thematic analysis. 79udy 1: Why are defendants pleading guilty?First, data from Study 1 was analysed to gain insight into the reasons that defendants are pleading guilty.Seven key considerations important to defendants when deciding whether to plead guilty were identified.These considerations are displayed in Table 1 in order of how frequently they were mentioned (note that considerations mentioned fewer than four times are not included). 80tudy 1 thus highlighted a range of reasons that defendants may choose to plead guilty in the current system.Two key themes in particular suggested ways in which autonomy might be being depleted through (i) the invocation of competing second-order values likely to challenge the ability of defendants to act in accordance with second order values relevant to the 'core' plea decision, or (ii) through undermining defendant feelings of self-trust, self-respect, and self-esteem and thus restricting their ability to act in accordance with relevant second-order values.These were: (i) Sentence reductions particularly when custody is involved.(ii) The time and cost involved in going to trial when compared to pleading guilty (including time spent on remand).
The two selected considerations are not intended to represent all possible restrictions on autonomy, but two that are likely to be important.Data from Study 2 was examined to explore practitioner perspectives on each of these areas. 81

Study 2: Exploring sentence reductions
The legal professionals interviewed in Study 2 noted positive aspects of sentence discounts.Lawyers described how they focused the mind of defendants and had a role in encouraging those who committed an offence to admit it earlier thus benefitting them in terms of a lesser sentence.It also allowed lawyers to move on to other work, freed the court's time and spared any victim(s) the stress of awaiting trial.There was, however, a widespread recognition that there were concomitantly problems with sentence discounts, particularly in cases involving reducing custodial sentences or avoiding custodial sentences all together.For example, considering the reduction of custodial time, BS1 stated: 'With prison sentences of several years, the pressure was said to be immense for defendants 79 V. Braun and V. Clarke, 'Using thematic analysis in psychology' (2006) 3 Qualitative Research in Psychology 77, 79 ('a method for identifying, analysing and reporting patterns across a data set allowing the authors to draw out new insight for the current exploration').80 For more information on current thresholds see n 87 below.81 While enhanced vulnerability was clearly also important from an autonomy perspective, this was not considered further due to presenting a variety of discrete issues relating to internal capacity for autonomy, and having been considered in depth in previous work, see n 46 above.because the stakes were just so high -and the advantages of obtaining the discount patently obvious'.Lawyers frequently noted that the amount of reduction offered by the sentence discount principle was largely a factor only for cases with longer sentences attached to them.In the following quote, DS7 articulates a common assertion from the research: … it's not a factor in as many cases as you'd think it is.It's not normally the deciding factor for a client in whether they plead guilty or not. 82The more serious the case, it can become more relevant to a client.For instance in a large drug case.So, if you're looking at a starting point of a nine-year credit brings that down to six, if you plead guilty, and you'll serve half -three.So your nine-year sentence is really three.Maybe, with release on tag, two and a half years, call it.So when you're looking at nine years, and you can be out in two and a half years, that can focus the mind.But if you've got a common assault, a domestic violence common assault in the magistrates' court, clients are not going to be persuaded to plead guilty because of credit. 83It'll be other factors, like the quality of the evidence, is the complainer going to attend, how much of a chance have I got at trial?They'll be far more concerned about those things than credit.
For DS7, then, sentence discounts are often irrelevant in the mass of cases they deal with in the magistrates' court.Here, with sentences of six months of imprisonment or less, defendants have other priorities in mind when deciding whether or not to plead guilty.However, it was also noted that custody can still be important in cases involving shorter sentences, particularly because a guilty plea can make the difference between a custodial and a non-custodial sentence.For example, DS2 stated: The problem is where it's on the cusp.So if they run a trial and lose, they'll be looking at prison, but if they plead guilty at the earliest opportunity probably a community order, and that makes a massive difference between someone keeping their job and not.And sometimes, sometimes, I think some people would rather plead guilty to something they haven't done and get the guaranteed community order, than to fight something where the evidence is heavy that they know they've not done and maybe end up with a custodial sentence if they lose.
Defendants wanting to avoid custody at certain times of year such as Christmas, was also raised by the lawyers in this study to highlight a situation in which custody might be seen as even more of a threat.
Lawyers in Study 2 raised concerns about the nature of sentence reductions as a 'sliding scale.'Specifically, participants noted pressure that might arise from the sliding scale of sentence discounts described above, where the largest discounts 82 Note that this statement is consistent with the findings of the Sentencing Council interviews with defendants, which suggested that the likelihood of conviction at trial was the key determining factor in whether a defendant would enter a guilty plea, see Sentencing Council, Attitudes to Guilty Plea Sentence Reductions (2011) at https://www.sentencingcouncil.org.uk/wp-content/uploads/Attitudes_to_Guilty_Plea_Sentence_Reductions_web1.pdf(last accessed 16 July 2021).83 Credit refers to the sentence discount that can be obtained as a result of pleading guilty.are typically given when defendants plead guilty at the earliest opportunity, combined with problems with the disclosure regime.According to lawyers in the study, the sliding scale of discount available at each stage created a burden on defendants.This position is summed up by DS17: … the pressure is about pleading guilty at the very outset.I think that is a lot of pressure for us to say to somebody, 'Look, you could be going to custody for a very long time, here.We haven't got all the evidence to advise you, but do you want to plead guilty or not.'I think that's a lot of pressure for us to take to somebody who could be looking at a long time in custody, potentially … I think we should be able to explore all avenues before they commit to anything, so I think there's too much pressure on us to advise them straightaway.'Are you pleading guilty or are you not?' It's a big decision for them to make.
There was significant agreement amongst these lawyers that it was unfair to place such weight on decisions made at this early stage.DS3 acknowledged the anxiety that defendants can feel when facing such important decisions so early in the process: What do you need at the point to be able to say, 'Well, yeah alright.I know I hit somebody.But was I acting in self-defence?' Yeah, it's quite, I can imagine it being quite a daunting prospect, quite a scary prospect for people, that position with, 'When do I plead guilty?Should I plead guilty?'And a lot of pressure … is put on the legal team to say, 'Well should I?' And obviously you have to remind them, it was your decision, but this is where we are … But that, saying, 'Yes, I'm going to plead guilty,' is a big decision for people.
Part of the reason that making an early decision can be so difficult was reported as being due to problems around disclosure.BS7 described the impact that a failure to receive early disclosure can have on plea decision-making: You are advising a client knowing that time affects them, because that affects their credit, et cetera, but you don't have all the information at that point.I know it's not practical to have all the information to hand at the very first hearing otherwise the expense would be crazy so there has to be a balance met, but often you can be at the magistrates' court and you will, you will, you can almost literally just have a case summary.And that's it.So it makes your -and maybe some scanned documents -it makes your position very difficult to advise at that stage and ensure, if they are going to plead guilty the full credit will be retained when you don't necessarily know all the information.
Responses suggest that the lawyer is hamstrung in what they can advise, and the defendant is being expected to make this -potentially life-changing -decision without knowing the true nature and full extent of the case against them. 84 Although note that conflicts faced by lawyers are likely to be significantly less severe than those faced in the less tightly regulated United States plea bargaining system, see for example R.K. Helm et al, 'Limitations on the ability to negotiate justice: Attorney perspectives on guilt, innocence, and legal advice in the current plea system' (2018) 24 Psychology, Crime & Law 915.
While defendants do almost always know what they have done (or have not done),some may not know whether what they have done amounts to a criminal offence and even more may not know whether they have a viable legal defence.
In addition, defendants who know they have not committed an offence may plead guilty because of over-estimating the strength of the case against them and thus their chance of conviction at trial.Disclosure problems were commonly raised by these lawyers and must form a part of an understanding into how the sentence discount operates in practice within the contemporary criminal justice system.

Study 2: Exploring time and cost
Data from Study 2 confirmed the importance of time and finances for defendants making guilty plea decisions, identified in Study 1. Data suggests that some defendants will want to avoid the impracticality of a drawn-out case, as suggested by DS13: I've had clients that are insistent that they haven't committed an offence, that they've said they can't afford to go to trial, they can't afford it in terms of time or money to put that off for a couple of weeks or months.They're looking to plead just to get credit even though they don't accept what the case is against them, essentially.
Here the defendant makes the judgment that it is better to 'get it over with' quickly and try to move on. 85Importantly, this may be critical in the decisions of defendants who may be innocent and want to maintain their status as a law-abiding citizen, specifically where other vulnerabilities such as caregiving exacerbate the need to move on.Further data also supported Study 1 findings in suggesting specifically that financial pressures are likely to be important for defendants when deciding whether to plead.Again, some of the participants in Study 2 explained that those defendants who do not qualify for legal aid, on the means test, will often feel pressure to plead guilty as they cannot afford to defend themselves. 86DS9 outlined this external influence: Talking about access to justice, there has been no evaluation of legal aid criteria, whether you know, financial eligibility, for over ten years, so lots of people who are on low-level jobs don't qualify for legal aid.think in reality lots of them will just say, 'Well I'll plead guilty just to get it over and done with.' Especially in the magistrates … and if they're not looking at prison either, they will deal with the consequences.
Relatedly, Study 2 also suggested that pressures from legal aid cuts -in terms of the fees available for certain types of work 87 -might influence the support of lawyers advising clients making plea decisions.This data builds on arguments that have previously been made which suggest that for cases that will stay in the magistrates' court, there may be a risk of sentence discounts being used to help solicitors achieve a higher volume of cases through encouraging a mass of guilty pleas. 88The presence of this effect was suggested by DS6: 'Most firms are looking at getting it in and maximising the income and then quickly very often pleading them out.They're not interested in trying to prepare for a trial.'Identifying such an anti-trial worldview supports McConville and Marsh's view that rather than being utilised as a tool to bolster the autonomy of defendants, the defence lawyer has been co-opted into the system of 'state-induced guilty pleas.' 89

THREATS AND OPPORTUNITIES
This data provides insight into guilty plea decision-making in the current system from the perspective of lawyers.In some ways the data is consistent with Sentencing Council data in that it suggests that the probability of conviction at trial is important to defendants, and that in many cases sentencing discounts are operating as intended.The data suggests that in the majority of cases,particularly in less serious cases where one third discounts amount to relatively short absolute time periods, autonomy is not unacceptably compromised provided the plea does not change the type of sentence imposed.Thus, particularly in cases tried in the magistrates' courts, which represent the vast majority of cases, 90 utilising such discounts presents a way to influence decisions of defendants without eliminating the influence of their core values relating to what is good, fair, or desired.The evidence presented here suggests that using such discounts is an effective way to maintain reasonable levels of guilty pleas while also respecting defendant autonomy.However, the data also point to threats to defendant autonomy.Specifically, the data suggests that defendants are making plea decisions in circumstances in which incentives to plead are leading to plea decisions being determined by second-order values (for example relating to fear of custody or a desire to avoid lengthy and expensive proceedings) that are distinct from the secondorder values applicable to the 'core' plea decision.Plea decisions driven by these values rather than values relating to the 'core' plea decision cannot be seen as autonomous decisions to self-incriminate.In addition, the data suggests that defendants are not making decisions in conditions conducive to self-trust, selfrespect and self-esteem, creating a risk that defendants will not be motivated to retrieve and apply relevant second-order values.Below, these threats are identified and opportunities for addressing these threats are presented.The threats and opportunities are intended as a set of important recommendations, rather than conditions that will necessarily be sufficient to ensure autonomy, and should also be considered alongside work examining the situation of defendants with enhanced vulnerabilities. 91

The threat of custody
First, the analysis highlights the importance of custody or the threat of custody in plea decisions.This threat will be particularly important where a defendant will be likely to receive a custodial sentence if convicted at trial but can get a community sentence if they plead guilty.As one participant noted, this can make the difference between a person keeping their job or not and can lead to people pleading guilty when they are innocent.This incentive has the potential to invoke second-order values that compete with the second-order values relevant to the central plea decision.For example, when considering whether they would rather self-incriminate or attempt to prove their innocence a defendant may be driven by a strong desire to avoid taking responsibility for something they did not do.However, where they may receive a custodial sentence at trial and a community sentence by pleading guilty their decision may be dictated not by second-order values relating to the 'core' plea decision, but by the importance of not being taken away from family or not losing employment.A completely different set of second-order values becomes important as a result of the incentive, and these values have the potential to dictate the decision. 92he data presented in this paper suggest that a similar situation may arise when a one third discount from a long custodial sentence is given since the absolute value of the one third discount may represent a significant period of time that may feel life-changing for the defendant.Extra care must be taken to monitor incentives to plead in such cases, particularly if the amended sentence length may influence other aspects of a sentence (for example, rules relating to early release).In these cases, a fixed discount, for example of one-year or two-years, 91 n 43 above.92 For empirical work supporting this contention, see R. K. Helm, 'Cognition and incentives in plea decisions: Categorical differences in outcomes as the tipping point for innocent defendants' (in Press) Psychology, Public Policy, and Law.
would be more protective of autonomy by ensuring that plea and trial outcomes are sufficiently comparable to avoid new second-order values being invoked as the result of the discount.
In addition, a person facing the threat of custody is likely to feel intimidated and threatened, which is likely to undermine their feelings of self-trust, self-respect, self-esteem, motivation, and ultimately autonomy.This threat is, of course, necessary in serious cases where custody is the only option.However, it is not necessary in cases where a community sentence could be an appropriate penalty (evidenced by the fact that it is the penalty likely to be received by pleading guilty).The problem here could be easily remedied by removing the provision in the sentencing guidelines stating that the one third reduction in sentence in exchange for a guilty plea can alter sentence-type.While this provision may be seen as primarily rewarding a defendant who agrees to give up their right to trial, its practical effect may well be to pressure defendants, and particularly defendants facing enhanced vulnerabilities, to give up their right to trial.Removing this provision would also restore consistency with the custody threshold provision, according to which custody should only be used where no other sanction is appropriate. 93Of course, removing this incentive may reduce the number of defendants willing to plead guilty.However, if procedure cannot encourage guilty pleas without violating the rights of defendants then priority must be given to protecting rights.The correct balance to be struck here can likely only be reached through further empirical research and consultation.

Time and cost
Second, the analysis highlights the importance of informal incentives in plea decisions, specifically the opportunity that pleading guilty gives to avoid the high costs and time commitment involved in trial.As one participant noted, some defendants cannot afford to go to trial in terms of time or money and look to plead even though they do not accept the case against them.Again, this reason for pleading guilty is unlikely to reflect a defendant's own second-order values relevant to the 'core' plea decision.This threat is likely to be particularly prominent where a defendant is remanded in custody and can be released from custody by pleading guilty, or where a defendant does not qualify for legal aid and is having to pay expensive legal and court fees themselves.The autonomy of defendants may be increased in this area through minimising restrictions on freedom in defendants awaiting trial and by minimising the time and cost involved in the court process.For example, research suggests that remand is currently being used in cases where this may not be necessary, such as cases where the defendant does not have a fixed address. 94Research conducted by the charity Transform Justice showed that 10,776 defendants charged with summary offences were remanded in custody in 2017 and 58 per cent of these 93 See J. Roberts and L. Harris, 'Reconceptualising the custody threshold in England and Wales' (2017) 28 Criminal Law Forum 477 (arguing against the ability of a guilty plea to convert a custodial sentence into a community penalty).94 See for example n 6 above.
defendants did not get a custodial sentence at the end of their case. 95Consistently referring defendants to organisations who can potentially provide a fixed address and therefore facilitate the granting of bail in these circumstances may be effective.Increased use of house arrest may also facilitate more consistent granting of bail in all but the most extreme cases (for example where the defendant poses a risk of harm to themselves or others that cannot be alleviated through house arrest). 96So, for example, the exception of the right to bail for those who pose a flight risk 97 might be removed and (absent other aggravating factors) such defendants might be monitored through house arrest.These changes also have the potential to increase the ability of defendants to act autonomously since an external environment is likely to be more conducive to feelings of self-trust, self-respect, and self-esteem than the restrictive and often poor conditions in a remand environment. 98he cost involved in trial might be reduced by removing restrictions on legal aid to ensure that all defendants have access to an effective criminal defence.Providing a more comprehensive legal aid system has the potential to 'open the door' to full trial for those who want to exercise their right to a trial, by making trial financially accessible.As per the suggestion of the Fabian society, a Right to Justice Act could codify existing rights to justice and establish a new right for individuals to receive reasonable legal assistance without unaffordable costs. 99In addition, the recent move towards online justice may offer a promising avenue to explore in reducing the time and cost involved in trial.This possibility must be considered in light of emerging research and literature relating to risks and benefits of online trials, and precautions necessary to protect the parties to trial in that context. 100

The lawyer-client relationship and legal education
Third, the analysis highlights the difficulties that lawyers, the primary protection system for defendants, face when advising clients in the current system.101These difficulties are important since effective legal advice is key to ensuring autonomous decision-making. 102As one participant noted, under current funding arrangements lawyers do not have sufficient time to process all the cases that they need to effectively, leading to a necessity to process cases quickly.This lack of time is hugely important in light of the conception of autonomy presented in this paper since it highlights the potential for lawyers to be led to being interested in defendants entering guilty pleas, rather than preparing for a trial.It is unlikely that this pressure on lawyers is conducive to them providing an environment in which clients feel safe and empowered to decide whether going to trial or pleading guilty is the right decision for them.Relatedly, as noted above, data from other research suggests that clients from minority ethnic groups plead guilty less often103 partly due to lacking trust in the lawyers advising them (and in the criminal justice system as a whole). 104Conditions of fear and mistrust are highly unlikely to foster feelings of self-trust, self-respect and self-esteem or to be conducive to clear understanding, and thus are likely to deplete autonomy and draw defendants away from relevant second-order values.
Problems with the lawyer-client relationship that may impinge on autonomy should be closely reviewed.First, there is a need for review of any influence of financial disincentives for lawyers with regards to criminal legal aid lawyers' practice.If there is a problem, then the fee structure must be adjusted accordingly for the benefit of defendants.Second, the use of community intermediaries for those who do not feel comfortable interacting only with their defence lawyer may help to promote autonomy in those who plead as a result of a lack of trust in the system and may help to address related inequality. 105ourth, there is a need for improved public legal education. 106This need has already been noted by the Low Commission report on legal aid and social welfare advice.Such education should occur in schools and in lifelong learning, in order that people know their rights and feel empowered to exercise them. 107his education has the potential to be important in helping defendants to feel empowered, although alone it is unlikely to allow defendants to effectively and meaningfully participate in legal proceedings.In their report, the Low Commission view this legal education as part of a continuum of support from informal and formal information to general advice, to specialist advice to legal help and legal representation.Providing greater information on rights and processes around plea has the potential to facilitate greater knowledge not only relating directly to legal matters but also to social welfare.Education relating to social welfare is important, since many of the pressures to plead we have identified relate to the ability of defendants to deal with broader, non-legal issues.Greater provision of information on the nature and dynamics of the lawyer-client relationship, including what the client is entitled to do and what their lawyers will expect of them is also required.It is unreasonable to expect defendantsin high stress situations -to absorb and process the limited information available to them and feel empowered to exercise their autonomy.The knowledge and confidence gained through education is likely to be empowering and to lead to more robust self-trust, self-respect and self-esteem when making guilty plea decisions, thus safeguarding autonomy.Such an approach would also allow (future) defendants to effectively and meaningfully participate in the case against them. 108

Judicial scrutiny
Finally, and perhaps most importantly, due to the importance of autonomy in the guilty plea process it is important that trial courts and appeals courts take the concept of autonomy seriously when accepting a guilty plea or considering the safety of a conviction that has arisen as a result of a guilty plea.Trial courts could potentially strike guilty pleas entered for reasons indicative of depleted autonomy (decisions dictated by second-order values not relevant to the 'core' plea decision).Such an enquiry might be similar to the subjective enquiry into the defendant's decision-making process that can be conducted in the Canadian legal system but utilising the suggested standard (although such an approach would need to be balanced with a need to respect defendant choice). 109This enquiry should involve asking defendants specifically why they have chosen to plead guilty, and assessing autonomy in light of the reasons given.Appeal courts should consider defendant autonomy when considering the safety of convictions that have arisen as the result of a guilty plea.Their analysis must not simply ask whether a defendant had a choice in the absence of external threats but must engage in an active consideration of whether the defendant was able to make their decision in line with their own second-order values relating to the 'core' of the plea decision.This analysis may differ by defendant and should involve actively understanding the incentives to plead faced by defendants, and how these might have impinged on their autonomy.This type of individualised analysis of choice has been suggested by Jonathan Herring in the context of sexual offences, where Herring suggests that an inquiry into consent should analyse a defendant's justifications, the context, and the victim's whole story in order to understand the victim's values, relationships, and best interests. 110his enquiry would need to transform consideration of autonomy as an opportunity concept into consideration of autonomy as an exercise concept.It would be necessary to ask whether the defendant had sufficient autonomy in practice to legitimise their conviction.To answer this question, courts must consider the incentives to plead faced by the defendant and also the conditions they were in when making their decision to plead.Where autonomy has been undermined defendants should be eligible for a retrial in which they can contest their guilt and have their case decided on the basis of evidence.Without this meaningful enquiry into autonomy being conducted, a change in the standard of autonomy may be ineffective.This result can be seen in the Canadian system utilising the related concept of voluntariness.In that system, although a relatively robust definition of voluntariness is employed, empirical evidence suggests that judicial scrutiny of voluntariness is relatively limited.111

CONCLUSIONS
When a defendant agrees to give up the right to a fair trial, they not only give up that right but allow the state to impose punishment that curtails other fundamental rights.It is reasonable to say that defendants themselves should be able to avoid a full trial and accept their status as a convicted person if that is what they want to do, provided this decision is driven by their own values relating to that 'core' plea decision. 112However, the reality of the system is that individuals are giving up fundamental rights as a result of government action in circumstances where the government benefits from those rights being given up.As such, the guilty plea process must be carefully monitored as closely as, if not more closely than, the full trial process.It is clearly unacceptable to rely on a conception of autonomy that is essentially a legal fiction.
In order to protect defendants, it is necessary to ensure that to the maximum extent possible plea decision-makers are empowered to act in accordance with their own second-order values relevant to the 'core' plea decision (ie the decision to voluntarily self-incriminate rather than contest guilt).This suggested standard requires meaningful engagement with the concept of autonomy as an exercise rather than opportunity concept in individual cases.The proposed set of reforms have the potential to bolster defendant autonomy.These include eliminating the ability of a plea to influence sentence 'type,' reducing the impact on defendants of the time and cost involved in trial, including using any possible alternatives to remand, ensuring funding structures for legal professionals allow time to deal with cases in a way that leaves defendants feeling informed and empowered, providing public legal education with the potential to empower citizens in advance of being accused of a criminal offence, and encouraging judicial scrutiny of guilty plea decisions.
Implementing these reforms has the potential to begin to create an environment in which defendants can make decisions that much more closely reflect truly autonomous choices and thus an environment that more appropriately respects fundamental rights.Importantly, these reforms may also help to reduce inequality in the current system, through reducing the impact of pressures external to the 'core' plea decision that may influence different groups differently.However, there are also limitations to the proposals.First, the proposals would require an overturning of current trends in the criminal justice system, particularly in relation to legal aid cuts and funding of the criminal justice system more generally. 113Overturning such trends represents a significant challenge, which has led some commentators to describe the future of legal aid in England and W a l e sa s' b l e a k .' 114 Thus, proposals need to be considered in the context of the existing literature on the legal aid system and funding for criminal justice more generally, which provides further compelling evidence that reform in this area is needed. 115Second, no intervention alone is likely to be sufficient to avoid depletion of autonomy.The proposed reforms should form part of an ongoing dialogue between the state, professionals in the criminal justice system, and defendants, to find ways to facilitate autonomy throughout the prosecution process.Finally, it should be noted that implementation of the suggested proposals has the potential to lead to more cases being taken to full trial (although this is not necessarily the case since increased autonomy could actually increase guilty pleas in some contexts).Any reform with the potential to cause more cases to go to trial may be considered undesirable in the current system, with a large backlog of cases made worse by the coronavirus pandemic. 116owever, the suggestions are particularly important to consider in light of the current backlog, which may enhance pressures to plead, and should contribute to the ongoing debate about the future of the criminal justice system beyond the current pandemic.
Most immediately, comprehensive research is needed in order to fully understand defendant autonomy in the guilty plea process and to further guide and nuance suggestions for reform.It must go beyond assessing whether guilty pleas work for the majority of defendants, and instead dig into the values that are important in the plea decision-making process and potential depletions of autonomy that are likely to permeate the system and lead to non-autonomous waiving of fundamental rights.
63 G. Dworkin, The Theory and Practice of Autonomy (Cambridge: Cambridge University Press, 1988) 7. 64 For example, in the family law context see M. Fineman, The Autonomy Myth: A Theory of Dependency (New York, NY: The New Press, 2005).65 See, for example, J. Anderson, 'Disputing autonomy' (2008) 9 Sats-Nordic Journal of Philosophy 7. 66 Note, these values are not necessarily the same as what a defendant would ideally 'want'.So, aguilty person may well ideally want to get away with a crime they have committed but would likely recognise conviction would be reasonable given their offending.
Under this conception, threats have a special status in restricting 18 K.Atkins, 'Autonomy and the subjective character of experience' (2000) 17 Journal of Applied Philosophy 71, 74.

Autonomy as the ability to conform an act to a mental state that represents one's own point of view
-making, for example generating more guilty pleas through providing larger discounts.Defendant autonomy is threatened by this conflict of interest, necessitating close monitoring of decisions and related autonomy.
61 See,for example,W.Heffernan and John Kleinig (eds),From Social Justice to Criminal Justice:Poverty and the Administration of Criminal Law (Oxford: OUP, 2000).62See S.J. Morse, 'Against control tests for criminal responsibility' in Ferzan, Robinson and Garvey (eds), n 25 above, 449.decision

Table 1 .
Considerations identified as important to defendants when deciding whether to plead guilty And then they're being told, 'Well, you know, if you want to instruct me for a trial it's going to cost you x-thousands of pounds,' which they don't have.And then they've got no option but to defend themselves if they, if they do decide to plead not guilty and have the trial run.I 85 See L. Skinns, 'Let's get it over with: early findings on the factors affecting detainees' access to custodial legal advice' (2009) 19 Policing and Society 58; V. Kemp 'No time for a solicitor: implications for delays on the take-up of legal advice' (2013) Criminal Law Review 184.86 Most obviously when they cannot afford to pay for representation themselves.Research suggests that the merits test in the magistrates will leave many defendants unrepresented, see Transform Justice, Justice Denied?The Experience of Unrepresented Defendants in the Criminal Courts 2016 at https://www.transformjustice.org.uk/wp-content/uploads/2016/04/TJ-APRIL_Singles.pdf(last accessed 28 June 2021).
87 Defendants, to receive legal aid, must obtain a representation order and satisfy (i) the interests of justice test and (ii) the means test.Crown Court cases will always satisfy the former; at the magistrates' those charged with non-imprisonable offences will automatically fail the first test.Legal Aid Lawyers and the Quest for Justice (Oxford: Hart, 2013).89 n 48 above.90 See Ministry of Justice, Criminal Court Statistics Quarterly, England and Wales, January to March 2019 (2019) at https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/812556/ccsq-bulletin-q1-2019.pdf(last accessed 16 July 2021).