Guilty pleas in children: legitimacy, vulnerability, and the need for increased protection

This article assesses the extent to which current guilty plea procedure is consistent with legitimations of criminal convictions, with a focus on decision making in child defendants. I argue that in the context of plea decisions in children, the criminal justice system must ensure that defendants make decisions that result in accurate convictions that are reached in a fair way that respects rights. The current system does not do this due to an almost exclusive focus on autonomy. This focus is likely to be leading to illegitimate convictions, most importantly children pleading guilty when innocent. Drawing on psychological theory, I develop a model of guilty plea decision making and draw on this model to identify relevant vulnerabilities of child defendants. Based on this analysis, I identify ways in which current procedure in England and Wales may be leading to systematic problems with the legitimacy of convictions of children, and suggest reforms to enhance such legitimacy. These reform suggestions focus on England and Wales but have implications for plea systems around the world.


INTRODUCTION
Many criminal justice systems around the world are heavily reliant on guilty pleas. 1 This reliance is evident in England and Wales, where the majority of defendants plead guilty rather than contest their guilt at trial, and the system is dependent on them doing so. 2 In the first quarter of 2019, 66 per cent of defendants entered a guilty plea to all counts against them in the Crown Court, 3 and calculations using data from 2016/2017 show that in that year the Crown Prosecution Service (CPS) obtained 91.65 per cent of its convictions via plea. 4A relatively high reliance on guilty pleas can also be seen when looking specifically at child defendants (defined as defendants under the age of 18).In 2019, 61 per cent of child defendants in the Crown Court pleaded guilty (with 58 per cent pleading guilty at their first hearing), and 47 per cent of child defendants in the youth court pleaded guilty at their first hearing. 5his reliance on guilty pleas is indicative of a criminal justice system that revolves around selfincrimination in addition to trial by judges and juries. 6In this context, the legitimacy of a very significant number of convictions is dependent on the decision making of defendants themselves.In the cases of the approximately 20,000 child defendants convicted and sentenced each year, these decisions are made by children. 7While young children in particular are likely to be reliant on advice from parents, guardians, or legal representatives, the ultimate decision to plead guilty (or not guilty) rests with the children themselves.
Importantly, plea decisions are not merely decisions to admit guilt.Research on guilty pleas generally suggests that in the current system innocent as well as guilty defendants plead guilty, and that decisions to plead guilty are often made as the result of complex considerations and pressures. 8This complexity is partly due to the fact that the system actively incentivizes people to self-incriminate through pleading guilty, and to do so as early in the prosecution process as possible.Sentence reductions for child defendants mirror those for adult defendants but are found in the Sentencing Council's definitive guideline on Sentencing Children and Young People. 9This guideline provides for a one-third reduction in sentence (subject to exceptions) where a defendant enters a guilty plea at the earliest possible opportunity.This reduction is lessened the further along in proceedings that a defendant pleads guilty, down to a maximum of a one-tenth reduction in sentence for defendants who plead guilty on the first day of trial.This discount can result in the imposition of a different type of sentence by pleading guilty.In the case of child defendants, pleading guilty can result in the imposition of a referral order or a youth rehabilitation order rather than a custodial sentence (typically a detention and training order).
In this system, children -whom the law recognizes as being too immature to vote, to drink alcohol, or to gamble -are making complex decisions to incriminate themselves, which may influence the rest of their lives. 10Despite the importance of these decisions, research in England and Wales has not examined guilty pleas in children (and in fact almost no research in this area has been done outside North America), and no published data provides insight into such pleas. 11Understanding and monitoring guilty pleas in children is vital, because children have immature cognitive, social, and neurobiological systems that influence their decision making. 12Psychological research, supported by accumulating plea-specific research in the US context (which involves a different plea system but many of the same underlying psychological constructs), suggests that as a result of these immaturities, children are more susceptible to pressures to plead guilty, 13 and also more likely than adults to plead guilty when innocent (meaning when they have not in fact committed the crime to which they are pleading guilty). 14This finding is also supported by a significant body of research across multiple jurisdictions showing that children have a particular at 108-112 (including a summary of published literature pertaining to juvenile guilty pleas, which shows that all relevant work to date has been carried out in the US or Canada).However, note that early empirical data collected by the author confirms the likely complexity of guilty plea decisions in children in the England and Wales context, and the relevance of existing research from North America for England and Wales.In survey data collected from 19 lawyers with experience of working with child defendants in England and Wales, participants were asked how important a variety of factors were to child defendants (on a scale from 0 to 100) when deciding whether to plead guilty.The most highly rated considerations (in order of rated importance) were advice from their own lawyer (mean rating 65.78), the comparatively better outcome at plea than if convicted at trial (mean rating 64.61), fear of sentence if convicted (mean rating 62.86), probability of conviction at trial (mean rating 57.78), short-term benefits (mean rating 54.53), and pressure from family members (mean rating 50.44).All of these factors were rated as more important than factual guilt (mean rating 48.89 developmental susceptibility to false confession more generally. 15Research examining the psychological mechanisms underlying guilty plea decision making also suggests that children may plead guilty on the basis of relatively small sentence reductions, and when this does not truly reflect their underlying values. 16This presents a real threat to children in the criminal justice process. 17Thus, developmental vulnerabilities must be accounted for when designing and evaluating guilty plea practice and procedure for child defendants and when monitoring compliance with relevant practice directions.As noted by the European Court of Human Rights (ECtHR) in the case of T v. United Kingdom (a case involving very young child defendants in the context of effective participation), 'it is essential that a child charged with an offence is dealt with in a manner which takes full account of his age, level of maturity, and intellectual and emotional capabilities'. 18n the next part of this article, I consider the legitimations underlying criminal convictionsaccuracy and fairness and respect for rights -to provide a benchmark against which to evaluate guilty plea decision making in children.I argue that even when convictions occur via guilty plea, the state has a duty to facilitate decision making that results in convictions that are legitimate according to these criteria, particularly in children.I draw on literature relating to the psychology of decision making to describe how the structure of choices faced by defendants, in addition to their underlying characteristics, can influence decisions, and how psychological research can inform procedure that maximizes the extent to which defendant decision making results in convictions that are accurate, fair, and rights compliant.In the case of child defendants, this requires an understanding of the psychology underlying guilty plea decision making generally, and the specific developmental vulnerabilities of children.
I then develop a model of guilty plea decision making based on fuzzy-trace theory, a psychological theory of memory and decision making that has been utilized in applied contexts, including legal and medical decision making. 19I draw on this model to identify specific vulnerabilities of child defendants that may lead to decision making that results in illegitimate convictions, most importantly pleading guilty when factually innocent.
In the final part of the article, I apply this theory to current procedure in the youth justice system (YJS) in England and Wales and assess the extent to which current protections for young defendants are likely to address identified vulnerabilities.I conclude that current protections fail to address a number of important developmental vulnerabilities, and I make suggestions for changes in practice and procedure that have the potential to increase the legitimacy of child convictions via guilty plea.

LEGITIMACY OF CONVICTIONS IN A GUILTY PLEA SYSTEM
As practice in the criminal justice system now largely revolves around the decision making of defendants, it is important to understand this decision making and to ensure that decisions are being made in a way that results in legitimate convictions.Although the criminal justice system includes safeguards to protect innocent defendants from pleading guilty (including pre-trial mechanisms for the scrutiny and weeding out of evidentially weak cases, and independent legal advice), defendant decision making itself remains vitally important since it is defendant plea decisions that determine whether or not the prosecution is required to prove the case against the accused at trial.Criminal convictions demand the highest levels of legitimation for several primary reasons.First, conviction and punishment are typically harmful interferences by the state in the life of a citizen.Conviction of a criminal offence involves the curtailment of fundamental rights, including those guaranteed by the European Convention on Human Rights, such as the right to liberty and the right to respect for family and private life.Second, criminal punishment is costly to the state, and so should be justified.Third, criminal conviction involves the imposition of community condemnation and stigma. 20uch convictions are legitimized by being both accurate and obtained in a fair and rightscompliant way (in accordance with principles of distributive and procedural justice). 21These legitimations are based in part on respect for the rights of individual defendants, most notably the idea that the state should minimize harmful interference in citizens' lives and that the innocent should be protected from conviction.In this context, being wrongfully convicted is a deep injustice and substantial moral harm. 22Since accuracy and fairness and respect for rights are essential to the legitimacy of convictions, these are recognized as crucial normative goals of the criminal justice system, 23 although their relative importance is debated. 24The importance of these goals is recognized in Section 1.1 of the Criminal Procedure Rules, which state that (1) The overriding objective of this procedural code is that criminal cases be dealt with justly.Such legitimations are also important in reflecting both retributivist and consequentialist justifications underlying the imposition of punishment on an individual (although accounts are based on distinct rationales, both are best served where convictions are accurate and fair). 26he legitimacy of convictions (through accuracy and fairness and respect for rights) is promoted by regulations surrounding trial by magistrates or jury.For example, in the majority of judgments the finder of fact is required to be convinced of the defendant's guilt 'beyond a reasonable doubt', 27 and the evidence presented to finders of fact is regulated to facilitate accurate and unprejudiced determination of guilt. 28However, the furtherance of these goals is far less clear in the case of convictions obtained by guilty plea, where there is very little regulation of defendant decisions, beyond the fact that they should be made autonomously (in other words, voluntarily). 29The notion of autonomy utilized when regulating pleas is relatively narrow, focusing specifically on the plea decision itself being free from improper pressure.For example, in Natsvlishvili and Togonidze v. Georgia, the ECtHR describe it as essential that 'the defendant's plea must always be made voluntarily and free from improper pressure', 30 and in R v. Turner, Lord Parker CJ emphasizes the requirement that the defendant should have 'a free choice . . . to plead guilty or not guilty'. 31his conception of autonomy is distinct from the broader notion of autonomy, in terms of selfgovernance more generally, that might require a conviction to be accurate (since self-governance is undermined when an innocent individual is convicted). 32nsuring that plea decisions are made voluntarily is clearly important, and is the focus of rightsbased restrictions on plea procedure. 33However, it does not safeguard the accuracy or fairness of convictions and should not, by itself, be regarded as legitimizing convictions, particularly those of children.This is in part because the autonomy of defendants in the criminal justice system is highly compromised.Defendants are in an inherently vulnerable position with their autonomy and rights restricted (such as through remand in custody and through bail conditions).This means that the autonomy of the defendant is severely constrained by the situation in which they are placed and the choices with which they are presented (something over which they may have no control).In some circumstances, the mere presentation of the choice can be coercive even where the ultimate decision is made 'autonomously' in the narrow sense described above.To give an extreme example, in the US plea bargaining system, research suggests that defendants can face extreme discrepancies between potential trial outcomes and plea outcomes, such as the possibility of decades in prison if convicted at trial but a year or two in prison if convicted by guilty plea, 34 or the possibility of the death penalty if convicted at trial but not if convicted by guilty plea. 35Faced with this choice, an innocent defendant may well 'autonomously' decide to plead guilty. 36utonomy alone can also not ensure fair outcomes.For example, at least in the current system, different racial groups appear to react differently to incentives to plead guilty and thus make systematically different plea decisions.Research suggests that the overrepresentation of black males in the criminal justice system may be caused in part by their greater tendency to plead not guilty (and therefore not benefit from possible reductions to a non-custodial sentence). 37This tendency has been attributed to a lack of trust in the criminal justice system among black and ethnic minority communities, 38 and suggests that the current procedure is not fair from a distributive justice standpoint.
Relying purely on autonomy is particularly problematic in the case of children (and other vulnerable groups), for whom we typically accept that the exercise of autonomy should be restricted based on capacity.Children (under 18s) are not considered to have sufficient autonomy to decide whether to smoke, gamble, or buy fireworks, or to decide who to vote for in an election.This is consistent with a theory-based approach, according to which children may not be considered fully autonomous rights holders, but rather as individuals with agency, gathering and developing the assets necessary for full autonomy. 39herefore, when evaluating guilty plea practice and procedure, particularly for children, it is necessary to go beyond examining whether pleas are entered autonomously and instead to focus more holistically on whether guilty pleas result in accurate convictions reached in a fair way that respects rights.This is consistent with vulnerability theory, according to which the state has an obligation to mitigate human vulnerability and ensure that institutions are functioning in a fair way. 40n order to determine whether guilty pleas are likely to result in legitimate convictions, we must look at the interaction between the characteristics of defendants making plea decisions and the choices that are presented to those defendants.This is because research in cognitive psychology demonstrates that the way in which people make decisions is influenced both by their own characteristics (such as their preference for risk taking or their reward sensitivity) and by the choices presented to them (such as the degree of risk involved). 41The way in which characteristics interact with choices, and the way in which altering choices tends to influence decisions, varies systematically with age. 42As a result, the assumptions made by policy makers and lawyers about how defendants are likely to make plea decisions and the effects that incentives will have on them are particularly susceptible to error with respect to decisions made by children.Reliance on such assumptions therefore has the potential to leave children making decisions that they are cognitively pre-disposed to make inappropriately -for example, because incentives to plead guilty may lead to a systematic preference for pleading guilty even among innocent children.
Therefore, in order to maximize the legitimacy of convictions, in addition to providing appropriate support for children based on their developmental vulnerabilities, the criminal justice system must present children with plea decision options that are likely to result in accurate, fair, and rights-compliant convictions in children specifically.To do this, regulations governing guilty plea choices in child defendants should be based on (1) an empirically informed model of guilty plea decision making, and (2) an understanding of how developmental immaturity is likely to influence this decision making.

A model of guilty plea decision making
Providing a model to explain the cognition underlying guilty plea decisions is important in understanding the plea process.Such a model can provide a starting point that can then be drawn on to identify vulnerabilities in children that may lead to illegitimate convictions.When making guilty plea decisions, defendants are picking between a certain conviction with a more lenient penalty (which can often be confirmed in Crown Court cases through what is known as a Goodyear hearing) 43 and an uncertain conviction with a likely more severe (but usually unknown) penalty.In this way, plea decisions are similar to many 'risky' decisions that have been studied extensively in the psychological literature, such as the well-known 'dread disease problem'. 44Like plea decisions, these decisions involve picking between a sure outcome (such as winning $50, or saving 50 lives) and a risky outcome (such as a chance of winning $100 and a chance of winning nothing, or a chance of saving 100 lives and a chance of saving none). 45esearch on such decisions has confirmed that individual differences in decision makers, as well as the details of the decisions involved, can influence decisions to pick a sure outcome or a risky outcome. 46Research has also demonstrated illogical inconsistencies in risky decisions.Specifically, people prefer the risky option over the sure option when both options are framed in terms of losses and prefer the sure option over the risky option when both options are framed in terms of gains. 47his research provides an insight into the psychology of risky choice, which is applicable to guilty plea decision making. 48In fact, empirical research in the US context confirms that many of the same factors shown to be important in risky decision making are likely to be important in plea decisions.Research has shown that plea decisions are influenced by the preferences of a particular defendant (such as their risk preferences), 49 the sentence discount that they can obtain by pleading guilty compared to the likely outcomes at trial, 50 the probability of conviction at trial, 51 and the framing of outcomes as gains or losses. 52Research has also highlighted additional influences that are important specifically in the guilty plea context, such as factual guilt 53 and pressures from lawyers. 54he prevailing model of guilty plea decision making in psychology, economics, and law is the 'shadow of trial' model. 55According to this model, a defendant will consider the expected value (the possible outcome multiplied by the approximate probability of it occurring) of a guilty plea compared to the expected value of going to trial when deciding whether to plead guilty.If a defendant is risk neutral, meaning that they do not have a preference for either seeking or avoiding risk, they are expected to plead guilty when the expected value of the plea (the certain punishment) is less than the expected value of trial (the approximate probability of conviction at trial multiplied by the likely punishment if convicted).However, this model is based on outdated psychological models of decision making relating to risk (primarily expected utility theory). 56More modern theories have built on this model through examining the cognitive mechanisms involved in decision making relating to risk.These theories have the potential to more fully explain guilty plea F I G U R E 1 A representation of the guilty plea decision-making process according to FTT NB These stages are not necessarily entirely sequential and can occur in parallel to some extent.decision making. 57One theory that has been specifically applied to guilty plea decision making is fuzzy-trace theory (FTT). 58FTT is a dual-process theory of memory and decision making that predicts and explains observed behaviour in studies of risky decision making and that has been formalized in mathematical models of memory and decision making. 59ccording to FTT, decision making is made up of four types of processes: encoding information relating to decision options, retrieving information relating to decision options, retrieving relevant values, and applying those values to the decision options to reach a decision. 60Figure 1 illustrates these processes in the context of guilty plea decision making.Decision making is influenced by the way in which each of these processes occurs, which is separable from other influences that have been traditionally recognized as important in decision making, including time, social pressure, and reward sensitivity and inhibition (although note that these influences may interact with the decision processes). 61n addition to these factors, FTT posits that decision making is influenced by the way in which information involved in a decision is represented and processed.When an individual processes information (such as the consequences associated with pleading guilty), two representations of the information -gist and verbatim -are stored simultaneously and separately.Gist representations are the bottom-line meanings that a person extracts from information such as simple qualitative representations of numbers.Verbatim representations are detailed superficial representations of exact information, including precise numbers and wording.For example, when a defendant is told that they are likely to receive a six-month probation sentence at trial, they will encode the surface form of this information (six months of probation) but also the gist (which may be something like a short period of a relatively non-onerous/non-harmful sentence).Although both gist and verbatim representations are encoded, when making decisions, individuals will tend to rely on either more gist-based representations or more verbatim-based representations.Individuals relying on gist representations process information in a fuzzy and impressionistic way (gist processing), while individuals relying on verbatim representations process information in a more precise and superficial way (verbatim processing).
When making decisions, a neurotypical adult (an adult demonstrating typical cognitive development) will generally rely on gist representations more than verbatim representations.This means that where possible they will reason categorically (in other words, based on meaningful categorical distinctions such as 'a risk of a custodial sentence v. no risk of a custodial sentence' or 'admitting to something that I did not do v. maintaining innocence at trial'), rather than conducting precise trade-offs of risks and rewards, even where precise information has been presented to them (such as the exact probability of conviction at trial and the exact outcomes if convicted at trial, as predicted by the shadow of trial model). 62This processing is important when comparing options to one another.Take the example of a defendant who is told that they are likely to receive a six-month sentence of probation at trial.Imagine that the defendant is told that by pleading guilty they can reduce their sentence by a third, to four months.A defendant relying on verbatim representations will then be directly comparing four months and six months, meaning that the two-month reduction will be appealing and could lead to a plea.In contrast, a defendant relying on gist (absent any specific meaning-based significance of the two-month reduction for them, such as not missing an important work start date) will likely be comparing a short period of a relatively non-onerous/non-harmful sentence to another short period of a relatively nononerous/non-harmful sentence (since both options have the same gist despite being different on a verbatim level).Such a defendant would be far less likely to decide to plead guilty on the basis of this sentence reduction.Existing research across several fields demonstrates how reliance on gist can improve decision making in this way. 63n addition to resulting in less precise and more categorical decision making, reliance on gist tends to lead to decision making that reflects values.This relationship between gist and values occurs because values are typically stored in long-term memory as 'gisty' representations, and are more easily cued when a person relies on gist representations of options (this is due to their similarity, a well-known principle of retrieval cuing). 64As a result, values (such as 'under most circumstances I should not admit doing something I didn't do') are reflected in plea decisions in addition to likelihood of conviction and outcomes at trial. 65The predicted influence of values in plea decisions is supported by experimental research that shows that factual guilt or innocence predicts plea decision making even when controlling for the probability of conviction. 66Thus, reliance on gist is beneficial when making decisions relating to risk since it is influenced by meaningful but not superficial distinctions and because it leads to appropriate retrieval and application of values.

Developmental psychology and plea decision making
Developmental differences in decision making can occur at each stage of the plea decision-making process.Four aspects of typical cognition in children make it likely that, without additional support, they will be particularly susceptible to making plea decisions that result in convictions that are not legitimate: differences in encoding and retrieving accurate information; differences in mental representations retrieved and subsequent cognitive processing; differences in inhibition, reward sensitivity, and the influence of short-term benefits; and differences in susceptibility to pressure. 67In this analysis, I focus specifically on general psychological vulnerabilities of children.However, these vulnerabilities should be considered in the context of the enhanced vulnerabilities of many children appearing in the YJS.For example, many children in the system come from difficult family backgrounds where drug and alcohol misuse, physical and emotional abuse, and offending are common. 682.1 Difficulties in encoding and retrieving accurate information misinformed plea decisions.Research in the US context has empirically demonstrated difficulties that children may face with understanding when making plea decisions.69 As discussed below, some deficits in understanding are recognized and addressed by current legal procedure.70 However, deficits in communication make it difficult to accurately assess understanding.71 For example, children often have response bias, which leads to a tendency to answer 'yes' or 'no', rather than admitting that they do not know or require clarification.72 Similarly, children may falsely affirm statements made by adults due to feeling social pressure or presuming background knowledge of the interviewer.73 In a 2014 report, it was noted that over 60 per cent of offenders in the youth court had communication difficulties, of whom around half had poor or very poor communication skills.74 The same report also specifically noted the risk that children can appear to understand much more than they do.75 Research studies have also consistently confirmed deficits in language and literacy in child offenders. 76or example, one study found that 57 per cent of child offenders in custody had reading levels below those of an 11 year old.77 Thus, communication difficulties are likely to be particularly problematic in child defendants.
This potential lack of understanding in children is important for the legitimacy of convictions in terms of accuracy and fairness and respect for rights (autonomy).First, it is important for accuracy since children need to understand an offence to know whether they committed it, and appreciating what it means to plead guilty and the implications of doing so are likely to protect innocent defendants from entering a guilty plea.Second, it is important for fairness, since accurate knowledge facilitates the ability of a defendant to make a decision that reflects their true underlying preferences.

3.2.2
Differences in mental representations retrieved and subsequent cognitive processing decision making. 78Thus, children are predicted to retrieve and rely on verbatim representations to a greater extent than adults.This reliance on verbatim results in decision making that is more like that envisioned by the shadow of trial model, based on what can be relatively superficial comparisons.As noted above, this can result in sub-optimal decision making generally, and in the plea context specifically.This is because by retrieving and relying on verbatim representations children are likely to (1) neglect important meaning-based distinctions that are not captured by verbatim representations, which are inherently superficial; (2) be influenced by relatively superficial aspects of decisions; and (3) fail to retrieve relevant values.In the example above, an innocent defendant is deciding between a relatively low chance of conviction at trial, which would result in a six-month probation sentence, and a certain four-month probation sentence when pleading guilty.An individual relying on verbatim processing may decide to plead guilty (Option 2) through neglecting meaning-based information (such as the significance of having a criminal record at all and the importance of factual innocence) in favour of precise and superficial comparisons (such as four months v. six months) and through failing to retrieve values relating to factual guilt and innocence in the decision-making process (such as not wanting to plead guilty to a crime that was not committed).
Reliance on verbatim is important for a number of reasons.First, it means that child defendants are more likely to plead guilty to crimes that they did not commit.This is because the distinction between factual guilt and innocence is a meaningful distinction that may not be captured by superficial verbatim representations.In addition, most people endorse the value that they would not want to plead guilty to a crime that they did not commit, but only those relying more on gist are likely to appropriately retrieve and apply this value.Empirical research in the US context provides support for the prediction that children are more likely to plead guilty to crimes that they did not commit.When interviewed following the entering of a guilty plea, child defendants are more likely than adults to claim that they pleaded guilty when innocent. 79In addition, experimental work using hypothetical plea paradigms suggests both that child defendants are more likely to plead guilty when innocent and that values related to guilt and innocence (such as 'I would not plead guilty to a crime that I did not commit') are less likely to be reflected in the decisions of child defendants. 80Specifically, research found that under 18s actually endorsed the principle 'I would not plead guilty to a crime that I did not commit' more strongly than adults, but that this endorsement was not reflected in their plea decision making. 81econd, it means that children are more likely than adults to plead guilty based on what may be considered modest and reasonable sentence reductions.This is because although such reductions may not alter plea options in a meaningful way (and thus not influence those relying on gist), they may alter superficial comparisons involved in the plea decision.A study applying FTT to plea decisions in the US context found that even a relatively minor sentence reduction involving probation rather than custody (two years of probation to one year of probation) was enough to 78  entice child 'defendants' (including innocent child 'defendants') into pleading guilty in a vignettebased study. 82

3.2.3
Differences in inhibition, reward sensitivity, and the influence of short-term benefits Children, particularly during adolescence, have been shown to have generally low levels of inhibition and high levels of reward sensitivity.Neurodevelopmental theory suggests that during adolescence, there is an imbalance between 'hot' motivational affective systems and 'cold' deliberation and inhibition.Specifically, brain regions implicated in cognitive control (prefrontal cortical regions) are thought to develop linearly with age, while affective brain regions (subcortical areas) develop faster and are hyper-responsive during adolescence. 83As a result, children between the ages of 12 and 17 have comparatively high levels of reward sensitivity, and insufficient inhibition to control decision making.This more affective decision making has been demonstrated in other contexts; for example, research suggests that young smokers give little or no conscious thought to the risks of smoking and are instead driven by affective impulses. 84This type of reasoning underlies age limits placed on many activities, including drinking alcohol and smoking. 85n the context of plea decision making, the imbalance between reward sensitivity and inhibition is likely to influence the information retrieved about decision options, and the values retrieved and applied as part of the decision-making process.For example, when making decisions, children may retrieve information about immediately rewarding consequences of decision options (which often relate to short-term benefits, such as getting a trial over with) and neglect information relating to important long-term consequences (such as the effects of a conviction on career prospects). 86This effect is interesting given children's objectively longer time horizon, but is consistent with work suggesting that adolescents in particular struggle to envision and plan for consequences that are not immediate. 87This over-influence of short-term consequences has been demonstrated in empirical work.Specifically, the authors of a study in the US context involving interviews with defendants concluded that child defendants appeared to be overly influenced by short-term benefits. 88Importantly, the research showed that reasons identified for pleading guilty often (60 per cent of the time) reflected a short-term orientation, whereas reasons identified for 82 Id. pleading not guilty often (69 per cent of the time) reflected a long-term orientation. 89This effect has also been demonstrated in the false confessions literature. 90his evidence suggests that increased short-term orientation in child defendants is likely to lead to increased guilty pleas.Research indicates that this is happening in practice, and short-term benefits rather than the case against them may be most important in child pleas.For example, one study found that evidence against them was not a significant predictor of plea decisions for defendants aged 11-14 (although note that it was for defendants aged 15-17). 91Child defendants are therefore at particular risk of making inaccurate guilty pleas where an immediate reward can be obtained.This finding is particularly important in light of the time discrepancy in getting to trial and completing trial when pleading guilty compared to contesting guilt.

3.2.4
Differences in susceptibility to pressure Finally, children are overly susceptible to pressure, including from the influence of others, in their decision making.Such pressure might exert a significant influence on the decisions of children.
The potential for pressure to interact with vulnerability in the legal context has been demonstrated with respect to false confessions, where research suggests that a key factor in unreliable confessions is the inability of a person to cope with pressure from the police. 92Other pressures have also been identified as important in this context. 93In the guilty plea context, research suggests that children are likely to be responsive to advice from parents, peers, and lawyers. 94While the influence of such advisors may be beneficial in some instances, this may not always be the case.Peers may, for example, pressure a defendant to plead guilty in order to protect a friend.
In addition, a tendency to comply with the recommendations of lawyers make child defendants particularly susceptible to any pressure to plead guilty placed on them by lawyers. 95This effect is particularly important where lawyers need to process many cases and may have incentives to deal with cases quickly through guilty pleas.This is thought to occur in the US plea bargaining system, which has been described as a 'meet 'em and plead 'em' system 96 in which defendants have very short time periods to make plea decisions. 97Research on guilty pleas in England and 89 Id.
Wales suggests that in some cases lawyers do pressure defendants to enter guilty pleas. 98This may be particularly harmful to child defendants.

YOUTH JUSTICE PRACTICE AND PROCEDURE IN ENGLAND AND WALES
It is important to consider practice and procedure in the YJS given the developmental differences addressed above and the need to maximize the legitimacy of convictions resulting from guilty plea.The current YJS in England and Wales 99 differs from the equivalent adult system and is already designed to address the needs and vulnerabilities of children (10-17 year olds), in various ways. 100The primary aim of the YJS is not punitive but to 'prevent offending by children and young persons'. 101Below, I consider the protections available for children in the current YJS and the extent to which they are appropriate for addressing the identified vulnerabilities, and make suggestions for reform designed to increase the legitimacy of convictions via guilty plea.These suggestions are not intended to be exhaustive.

Diversion from prosecution
In order to protect child defendants, and in recognition of the fact that crime committed by children is often committed impulsively, the criminal justice system as a whole has increasingly sought to deal informally with low-level offending by children.This informal treatment has involved diverting many children from the YJS, through diversion schemes.In 2018/2019, there were 5,179,038 police-recorded incidents of youth crime.The majority of these incidents were dealt with informally, with only 60,208 children being arrested and formally entering the YJS.Of these, 27,352 children were proceeded against at court and 8,552 cautions (formal warmings based on admissions of guilt) were given to children by the police. 102here are clear benefits of both diversion schemes and the use of cautions, and diversion schemes in particular have been shown to lead to better outcomes for children. 103When children are either diverted from the YJS or accept a caution, they are protected from ever having to make a guilty plea decision.However, these options may only be available to a child if they are willing to admit guilt. 104This is problematic as many of the same vulnerabilities that influence guilty plea decisions are likely to apply to these incentivized admissions.This is most problematic in the case of cautions, which form part of a child's criminal record. 105hile a full discussion of relevant procedure relating to cautions is beyond the scope of this paper, some of the suggestions made below relating to guilty pleas are also likely to be relevant to cautions.In the case of diversion schemes, removing requirements that children must admit guilt in order to enter the schemes has clear potential to more effectively protect children.One suggestion that has been made in this regard is that the more flexible criterion of 'accepting responsibility' might be used. 106However, such a requirement still implicates a child in criminal behaviour where they wish to participate in a diversion scheme.A requirement that children indicate that they would rather participate in criminal justice interventions than face prosecution may therefore be preferable.This does reduce what is sometimes thought of as a 'safeguard' to prevent undertaking criminal justice interventions with innocent children; however, since evidence suggests that children are willing to admit guilt when they have not committed a crime or do not know whether they have, this safeguard is unlikely to be sufficiently effective anyway.

4.2
The operation of the plea system Where children are proceeded against in court, cases start in the youth court (a type of magistrates' court specifically designed for children) and the majority remain there, with only a small number being referred to the Crown Court.In 2018/2019, 96 per cent of children sentenced were sentenced in the youth court, compared to 4 per cent in the Crown Court. 107The youth court is less formal than adult court and procedure is designed to make it easier for children to understand what is happening.In terms of support, children under 16 must attend court (in both the youth court and the Crown Court) with a parent or guardian and children aged 16-17 may attend with a parent, guardian, or someone to support them. 108Modifications in both the youth court and the Crown Court can also be made to ensure effective participation of child defendants.These include referring to children by their first name, taking time to explain court proceedings to them, avoiding complicated language, and using an intermediary to help with communication. 109he provisions for support are certainly likely to be helpful in facilitating understanding and comfort in child defendants.However, research suggests that many children still struggle with understanding even in the youth court. 110In addition, these provisions do little to address vulnerabilities that are not related to understanding.For example, the one-third sentence reduction for pleading guilty at the earliest opportunity continues to apply in the youth court, and so sentence length pressures as well as time and social pressures are likely to remain.It should also be noted for Justice Innovation, Who Should Be Eligible for Youth Diversion?Evidence and Practice Briefing (2019) 3, at <https:// justiceinnovation.org/sites/default/files/media/document/2019/Eligibility%20Criteria%20Briefing.pdf>. 105This can also be problematic in the case of diversion schemes, such as where community resolutions are imposed.
Although such resolutions do not constitute a criminal record, they are recorded on police information systems and taken into consideration if further offences are committed. 106Centre for Justice Innovation, op.cit., n. 104, p. 4. 107  that parental supervision may be less helpful for many young defendants who, as discussed above, often have difficult family circumstances.
Reform has the potential to improve the operation of the plea system for children through (1) tailoring sentence reduction guidelines, (2) removing Goodyear hearings, and (3) reducing time pressure.

4.2.1
Tailoring sentence reduction guidelines The incentives offered to children to encourage guilty pleas should be tailored to the child defendant population and based on an understanding of underlying developmental psychology.The current sentence reductions of between one-third and one-tenth of a sentence appear reasonable and modest.This has led to a presumption that such reductions would not lead innocent defendants to plead guilty. 111While this may be true in the case of neurotypical adults, such incentives may have a different effect on decision making in children.As noted above, children are thought to be more cognitively pre-disposed to make decisions in a relatively superficial, quantitative way, without significant impact of values relating to guilt.This decision making is similar to that anticipated by the shadow of trial model of plea decision making.Although, as noted above, this model does not explain many findings in the decision-making literature as a whole, it does relatively accurately capture the verbatim-style reasoning in which child defendants are thought to engage.
If plea decisions are based on something like expected value calculations and defendants can receive a one-third reduction in sentence in exchange for pleading guilty, defendants will plead guilty when they perceive their chances of being convicted at trial as medium to high (equivalent to an approximately 67 per cent probability or higher, although it is unlikely that defendants will consider specific numerical probabilities).This is because the expected value (sentence length) of trial in these cases (a 67 per cent chance of X = .67X)will be longer than the expected value (sentence length) of the plea (a 100 per cent chance of .66(X)= .66X).For example, imagine a defendant who faced a likely two-year (24-month) custodial sentence if convicted at trial and knew that their sentence if pleading guilty would be around one year and four months (16 months).Here an expected-value type calculation would lead that defendant to plead guilty due to a superficial comparison between the 16-month sentence if pleading guilty and the moderately high chance of being convicted at trial and receiving a 24-month custodial sentence.Although real plea decisions will be complicated by many additional factors, this is the type of decision making anticipated in child defendants.This decision making is problematic since it suggests that child defendants who are not highly likely to be convicted at trial may plead guilty based on the one-third reduction in sentence.
For these reasons, the one-third sentence reduction is likely to be too large to use as a starting point when dealing with children.One solution may be to award tailored reductions to children based on less prescriptive guidelines and individual case circumstances (including defendant age) in order to strike the correct balance between recognizing the responsibility of the child in admitting guilt in their particular circumstances and also avoiding the potential for coercive discounts.Less precise sentence discount recommendations would also be beneficial in avoiding expectedvalue type calculations and encouraging children to focus on the bigger picture.

4.2.2
Removing Goodyear hearings In the Crown Court, before entering a guilty plea, a defendant can request an indication of the maximum likely sentence that they will receive if they plead guilty, through a Goodyear hearing. 112Although such indications are only available in the Crown Court and therefore only influence a minority of child defendants, considering such indications carefully in the context of child defendants is important in protecting children accused of the most serious offences.Goodyear indications are likely to be problematic in cases of child defendants for two reasons.First, knowing the maximum sentence that will be handed down upon pleading guilty promotes the more quantitative and superficial decision making to which child defendants are already pre-disposed (even where a defendant does not know the sentence that they will receive at trial, they should have a reasonable idea of the sentence that they are facing if convicted).This type of decision making is problematic since it can lead child defendants to miss more important meaning-based distinctions.For example, the meaning-based impact of a criminal record may be less important to child defendants than a superficial comparison between community sentence lengths.Importantly, according to the research discussed above, this is likely to be the case even where a child understands the considerations involved, purely due to developmental cognitive disposition.As a result, even an explanation from a well-trained lawyer may not be sufficient to alter decision making.If child defendants do not know the exact maximum sentence that they will receive on pleading guilty (but have an accurate idea of an approximate maximum sentence), this prevents them from engaging in precise expected-value type calculations (through allowing them to disregard minutiae) and thus forces them back towards fuzzier, more qualitative decision making that is more likely to reflect important values including those relating to guilt.
In addition, Goodyear indications may lead to child defendants feeling pressured to plead guilty, since this can be seen as the tacit recommendation of the court.This pressure, on defendants of all ages, was the logic behind the prohibition on judges giving any indication of likely sentence in advance of a guilty plea prior to Goodyear.As the Court of Appeal noted, [a] statement that on a plea of guilty he would impose one sentence but that on conviction following a plea of not guilty he would impose a severer sentence is one which should never be made.This could be taken to be undue pressure on the accused, thus depriving him of that complete freedom of choice, which is essential. 113 noted above, children are especially likely to be susceptible to this kind of pressure.

4.2.3
Reducing time pressure The current sentence reductions given to defendants who plead guilty are intended to encourage guilty pleas to be entered as soon as possible in the trial process. 114This is done through providing for the largest discount if pleading guilty at the earliest opportunity, through to the smallest discount if pleading guilty on the first day of trial. 115In the context of vulnerable defendants more generally, Peay and Player have stated that this inducement to plead guilty at the earliest opportunity 'selectively ignores the vulnerabilities of specific groups of defendants and ratchets up their risk of making a false confession or receiving a more severe sentence'. 116In the context of child defendants, who are particularly vulnerable to pressure, this risk is especially important.In order to safeguard child defendants, this time pressure must be removed from the plea decision-making process.

Support from professionals
Children under 16 (or under 18 and in full-time education) automatically get legal aid for representation in court.Lawyers can provide important support to young people and could potentially mitigate the effects of some vulnerabilities.However, research has raised questions about the quality of representation in the youth court.Specifically, a youth proceedings advocacy review published in 2015 by the Bar Standards Board (BSB) and the Chartered Institute of Legal Executives found that (1) advocates were lacking in training in youth court procedures and sentencing, and (2) advocates often had difficulty in communicating and engaging with young defendants. 117This review noted the practice of youth courts being used as a 'training ground' for 'baby barristers' and being avoided by more able, experienced, and ambitious lawyers. 118hese findings led the BSB to bring in new standards for youth proceedings competencies. 119hese competencies are outlined in a 15-page document containing basic requirements for barristers representing children. 120In addition, in 2019 the Solicitors Regulation Authority (SRA) engaged in a consultation exercise to gather feedback on a proposal requiring youth court solicitors to qualify as higher court advocates (where acting as advocates in a case that would go to the Crown Court if brought against an adult).
While such standards may play a role in improving the quality of representation for children, they are unlikely to be sufficient in ensuring that lawyers representing children can effectively communicate with and represent children deciding whether to plead guilty.The BSB competencies do not address decision-making vulnerability, and while they provide some guidance on communication and engagement, this is relatively basic.SRA proposals may improve advocacy but are unlikely to improve lawyer understanding of and ability to confront child decision-making vulnerability.
The effectiveness of the support that children currently receive could be improved by ensuring that lawyers working with child defendants receive tailored training to help them identify deficits in understanding, appreciation, and decision making more generally that are important when a child defendant decides whether or not to plead guilty.In terms of understanding, training should be developed based on existing protocols for interviewing children.Although such protocols focus on extracting information from potential witnesses or the accused,121 they also have the potential to be important in probing the true understanding and appreciation of children.Training can provide guidance to lawyers on assessing whether child defendants have accurately encoded and retrieved information.For example, when seeking to clarify understanding, lawyers should be sure to ask open-ended questions (such as 'What are the consequences of pleading guilty?') rather than closed questions (such as 'Do you understand the consequences of pleading guilty?') and to use language that is appropriate and understandable for children.Through using such questions and language, lawyers can gain an informed understanding of the information that a child has encoded and is relying upon when making their decision. 122n terms of appreciation, lawyers should be alert to the possibility that children might sound perfectly logical and recite verbatim facts but might fail to appreciate and appropriately account for meaningful aspects of plea decisions.Training can be provided to assist lawyers in helping children to take account of this meaning more appropriately, such as by enabling child defendants to generate personally applicable 'take-home' messages. 123t is possible that the use of intermediaries with research-based training to work with child defendants when making guilty plea decisions could also be helpful in this regard.Specifically, lawyers could seek the help of intermediaries pre-trial.The intermediary, as a communication specialist, would have the ability to more accurately ascertain the understanding and appreciation of the child and to facilitate effective communication between the child and their lawyer. 124sing intermediaries to address communication difficulties in child defendants would require the automatic provision of intermediaries to child defendants.This suggestion is not new, and was discussed in a 2014 report on the operation and effectiveness of the youth court. 125Empirical research suggests that providing intermediaries for all child witnesses would be relatively inexpensive, but also notes how under-resourced the intermediary system currently is. 126This provision would require a reversal of the current trends towards restricting access to intermediaries even for children. 127

CONCLUSION
The decision making of defendants is important to the legitimacy of convictions in the guilty plea era.Child defendants, with their developmentally immature decision-making systems, require specific tailored protections to avoid systematic wrongful conviction.Importantly, developmental vulnerability means that children are likely to be systematically pleading guilty to crimes that they did not commit in predictable circumstances.Empirical research and insights from cognitive psychology have the potential to be useful in designing procedure that can protect child defendants while maintaining a relatively high guilty plea rate.Such procedure could be designed to ensure that the choices faced by child defendants encourage guilty pleas only in guilty defendants and produce fair and rights-compliant convictions.Until such procedure is designed, there is a high likelihood that a significant proportion of convictions resulting from guilty pleas in child defendants will not be legitimate.This must be taken seriously since it threatens the justifications underlying criminal punishment, and the criminal justice system as a whole.
While certain recommendations can be made based on current knowledge, the potential for more fundamental evidence-based reform is hampered by the lack of available data on guilty pleas in the youth court, the lack of any psychological research on guilty plea decision making in England and Wales, and the lack of any existing legal research or academic commentary relating specifically to guilty pleas in child defendants in England and Wales.This article is intended to start a vital conversation in this area.Future work should focus on obtaining and analysing both field and experimental data to provide additional insights into current procedure and to test proposals for reform.It must also be noted that the situation in England and Wales is far from unique in this regard.Despite the prevalence of guilty pleas, there is very little research on child guilty pleas and almost no research on child guilty pleas outside the North American context.Future work must address this deficit to protect child defendants and to ensure the legitimacy of convictions of children around the world.

A C K N O W L E D G E M E N T S
This research was supported in part by a UK Research and Innovation Fellowship and in part by an ESRC Impact Accelerator Award.The author would like to thank Andrew Ashworth and Rachel Gimson for their comments on earlier drafts of this manuscript.The author would also like to thank the anonymous reviewers for reading the article and for their very helpful comments.

( 2 )
Dealing with criminal cases justly includes -(a) acquitting the innocent and convicting the guilty; (b) dealing with the prosecution and the defence fairly; (c) recognising the rights of a defendant, particularly those under Article 6 of the European Convention on Human Rights . . .
et al., 'Age Differences in Affective Decision Making as Indexed by Performance on the Iowa Gambling Task' (2010) 46 Developmental Psychology 193.
). Complete data and details of methods are available from the author.science1184; S. J. Blakemore, 'The Developing Social Brain: Implications for Education' (2010) 65 Neuron 744; C. Geier and B. Luna, 'The Maturation of Incentive Processing and Cognitive Control' (2009) 93 Pharmacology, Biochemistry and Behavior 212; B. J. Casey et al., 'Breaking and Accelerating of the Adolescent Brain' (2011) 21 Research on Adolescence, 21.Note that some adults, such as those with intellectual deficits, may also suffer from vulnerabilities: see Peay and Player, op.cit., n. 8.
Decisions' (2016) 40 Law and Human Behavior 611; T. M. Zottoli et al., 'Plea Discounts, Time Pressures, and False Guilty Pleas in Youth and Adults Who Pleaded Guilty to Felonies in New York City' (2016) 22 Psychology, Public Policy, and Law Reyna and Ellis, op.cit., n. 42; Reyna et al., op.cit., n. 41; V. F. Reyna and F. Farley, 'Risk and Rationality in Adolescent Decision Making: Implications for Theory, Practice, and Public Policy' (2006) 7 Psychological Science in the Public Interest 1; P. A. Tun et al., 'Response Latencies for False Memories: Gist-Based Processes in Normal Ageing' (1998) 13 Psychology and Ageing 230; T. N. Odegard et al., 'Attention to Global Gist Processing Eliminates Age Effects in False Memories' (2008) 99 J. of Experimental Child Psychology 96.