Georgi Gardiner: Beyond Reasonable Doubt

This month we are talking to Georgi Gardiner. Georgi has finished her PhD in Philosophy last year and currently holds a Junior Research Fellowship at St. John’s College, Oxford.


Welcome to Legal-phi, Georgi! I know that most of your work is on epistemology, rather than on legal or moral philosophy. Before anything, could you tell us why did you become a philosopher? And why the focus on epistemology?

Georgi: It’s wonderful to be here! Thanks for interviewing me.

I have always been drawn to philosophical questions. I didn’t study philosophy formally before my undergraduate degree, but we touched on philosophical topics in literature, sociology, and those kinds of subjects at high school. I would wonder about these odd questions: If something has the same structure as a hydrogen atom, but interacts with other atoms differently, is it still hydrogen? Is there a problem with soma (from Huxley’s Brave New World)? Is romantic love like a mental disorder? I just kept thinking about those kinds of weird questions. In Britain you have to decide which subject to study when you apply to university. I didn’t really know what philosophy was, but I had a sense that it would suit me.

Another factor was that I didn’t want to narrow my fields of study, and I figured that philosophy was the broadest subject I could pick. Philosophy is an extremely broad umbrella, so perhaps I was right about that.   

Could you give us a rough idea of your work on epistemology?

Georgi: I work on a few different topics: the nature and value of understanding, some topics in social epistemology (such as group belief and testimony), applied and legal epistemology, epistemic value, virtue, and luck. I also work on meta-philosophy, particularly on the ‘function first’ approaches that stem from Edward Craig’s work on the genealogy of knowledge attributions.

There are unifying themes to my work. One theme is that we need to invoke the idea of ‘normalcy’ to illuminate some phenomena. And normalcy isn’t just a matter of statistical likelihood; it can’t be easily reduced. I think epistemologists are reluctant to invoke normalcy. This is perhaps because we lack prospects for a reductive analysis. But I think normalcy plays ineliminable roles, and so we shouldn’t shy away from positing it in our theories. Another theme is functional roles. I think we can illuminate phenomena by considering their role in larger systems. In general, I am interested in understanding, how things hang together or cohere.

What made you start working on legal and moral philosophy?

Georgi: There has been a burgeoning interest in social epistemology recently. And in particular there has been an ‘applied turn’. This appeals to me. I think the epistemology of society is very rich. A lot of young queer people are told ‘it’s just a phase’ when they come out, which is patronising. But it is also epistemically interesting, because it’s denying the youth has this self-knowledge. But few people who come out as queer later think of themselves as straight. The belief ‘I am queer’ seems reliable: if you believe it, you are highly likely to be right. By contrast, a lot of people think they are straight when they are younger, and then much later realise they are queer. That pattern seems relatively common. Believing you’re straight isn’t on such firm epistemic footing. So adults tell queer youths ‘it’s just a phase’, but if anything we should direct this possibility towards people who think of themselves as straight. That is the kind of belief that has a poor track record, and would benefit from more critical scrutiny!

I took a social epistemology seminar with Alvin Goldman in 2012. We looked at the epistemology of science, of democracy, and of criminal law, which piqued my interest. And then in 2013 I took a Philosophy of Law seminar with Doug Husak. One way of understanding the theme of the course was applying language, epistemology, mind, and metaphysics to law. So, for example, it examined the relationship between forgetting and the recklessness-negligence distinction, how to understand attempts and intentions in the law, and the nature and role of vagueness in law. That gave me a broader sense of the available topics.

At the end of Husak’s course we read Larry Laudan’s arguments that the standard of proof governing criminal conviction should be radically lowered. The argument was largely based on crime statistics—it contrasted the chances of false conviction for a violent crime with the chances of being violently victimised by someone who was previously falsely acquitted for a violent crime. Laudan paints a grim picture about crime: a lot of serial offenders, a high risk of victimisation. I thought his statistical reasoning was seriously flawed. And I wrote a paper in response. (My paper is called ‘In Defence of Reasonable Doubt’.) Engaging with that project made me appreciate how deeply applied the topic is. Laudan’s research could be picked up by the media, or by legal professionals. It could be used to justify longer sentences or more aggressive prosecutions. It could even lead to a lower standard of proof for conviction. And I hope my paper can help oppose the picture Laudan paints. 

Regarding your work in legal philosophy, let me start with a basic question: what is legal epistemology?

Georgi: That is a great question. Legal epistemology examines epistemological issues raised by legal practice. I can explain through some examples. Legal epistemology also looks at the connections between ordinary epistemic ideas and epistemic ideas in the law: perhaps the idea of ‘reasonable doubt’ can illuminate both knowledge and legal proof, for example.

As Susan Haack writes, the law is ‘up to its neck in epistemology.’ Factfinding is a central aim of legal systems; legal proceedings try to determine what happened. This raises questions about how confident factfinders need to be. How much conviction is required for conviction? Are legal burdens of proof, such as beyond reasonable doubt, a matter of having confidence in the litigated claims, or do they require more robust epistemological conditions to be met, such as epistemic virtue, appropriate weighing of evidence, or modal conditions such as safety? Can the legal standards of proof be quantified and (relatedly) what roles can statistical evidence play in satisfying the legal standards?

Evidence law has two foundational principles: (i) all relevant evidence is admissible, unless there is a specific reason for its exclusion and (ii) all irrelevant evidence is inadmissible. This raises questions about how to understand relevance. The United States’ Federal Rule of Evidence says that evidence is relevant if ‘it has a tendency to make a fact more or less probable than it would be without the evidence’. So this claim needs to be unpacked, epistemologically. And the exclusion rules are epistemologically rich. What reasons—moral, political, economic, practical, epistemic—justify excluding relevant evidence? What kinds of evidence should be excluded? How should the legal system weigh the competing considerations? Evidence might be probative yet lead to less reliable outcomes because it elicits bias in factfinders. Prior crime evidence, for example, might be relevant yet prejudice can cause jurors to overestimate the evidential significance of the evidence or to find guilt in part as perceived justice for the earlier crimes. So it might be relevant but overall lead to fewer accurate verdicts.

Prior crime evidence is usually excluded. But there are some exceptions. An important exception is prior crime evidence in prosecutions of sexual violence. The rationale is that prior crimes bear on the likelihood of guilt and can help secure a conviction. But some feminist legal epistemologists challenge the exception, arguing that making exceptions for sexual violence perpetuates misconceptions and misunderstandings about rape—rape myths. They argue that allowing prior crime evidence in cases of rape perpetuates the myth that rapes are only committed by unusual, abnormal men. This can hinder people’s capacity to recognise rape. This is an example legal epistemology invoking hermeneutical epistemic injustice.

It would be impossible to canvas all of the topics in legal epistemology. But here are some other illustrative examples. Youngjae Lee draws on recent work on peer disagreement to examine whether disagreement within a jury is consistent with proof beyond reasonable doubt. We can investigate whether juries can form group beliefs. Legal scholars and judges grapple with how to identify and oversee expert witnesses. Scientists might prove a toxin causes a medical condition in general, but legal institutions typically need to establish whether the toxin caused the condition in the plaintiff’s particular case. Legal epistemology can help understand how to bridge that gap. Legal epistemology can evaluate the practice of using profiling for sentencing, and can illuminate whether practical stakes affect epistemic verdicts.

So those are some examples of legal epistemology. It’s a highly interdisciplinary topic. It draws on psychology, forensic science, behavioural economics, cognitive science, criminology, statistics, sociology, and legal scholarship.

Given what you’ve said, it seems that discussions about the legal standards of proof are at the core of legal epistemology. Take the ‘proof beyond reasonable doubt’ standard, which is the standard for criminal conviction in several jurisdictions. In a forthcoming paper you oppose the view that we should understand this standard in a purely statistical way. Why is that?

Georgi: Yes, some theorists quantify ‘proof beyond reasonable doubt’ as a statistical likelihood that the defendant is guilty. Often it is glossed as about 95% confidence. Perhaps the easiest way to cast doubt on the idea that ‘proof beyond reasonable doubt’ can be understood as a statistical likelihood is through a pair of cases known as the prison yard cases.

In the first vignette one hundred prisoners exercise in the prison yard. Ninety-nine prisoners together initiate a premeditated attack on a guard. Security footage reveals one prisoner standing against the wall refusing to participate. The footage is grainy, and so doesn’t reveal which prisoner refused to participate. The prison officials arbitrarily select a prisoner, Tom, and prosecute him for assault.

In the second vignette one hundred prisoners exercise in the prison yard. Five prisoners together initiate a premeditated attack on a guard. A second security guard, standing at a substantial distance, believes that he saw that a particular prisoner, Tom, was one of the attackers. The prison officials charge Tom with assault.

Plausibly in the first case the ‘bare statistical evidence’ licenses a relatively high credence in the target claim (perhaps 0.99). The evidence renders guilt probabilistically likely. But nonetheless there would be something wrong with finding a guilty verdict on this evidence alone. In the second case, non-statistical, ‘individualised’ evidence is available. The overall likelihood of guilt given the evidence might well be lower, but it is the right kind of evidence to satisfy ‘proof beyond reasonable doubt’.

So the pair of cases suggest that bare statistical evidence is insufficient for satisfying the ‘beyond reasonable doubt’ standard, and that something more than high probability of truth is required. This kind of example is from Charles Nesson’s 1979 ‘Reasonable Doubt and Permissive Inferences: The Value of Complexity’. The structure of this pair of cases bears a striking resemblance to the lottery paradox in epistemology. Indeed, one of the questions I am working on is to what extent the prison yard case is akin to the lottery paradox.

The prison yard cases are often used to indicate the standards of proof cannot be quantified. In general, I think we should be wary of attempts to formalise reasoning. Formalising and quantifying can be powerful; it is great where it is appropriate. The danger is trying to formalise and quantify beyond where it is appropriate.

So how do you think we should understand such standard?

Georgi: This is a much harder question! So far my work has mostly been arguing against existing accounts: I have argued against Duncan Pritchard’s safety account, Enoch, Spectre, and Fisher’s sensitivity account, Judith Jarvis Thomson’s causal-relation account, and some of the ‘moral encroachment’ accounts of what epistemic conditions are required for a legal burden of proof. I discuss some of these in my essay ‘Legal Burdens of Proof and Statistical Evidence’.

Martin Smith develops a ‘normic support’ account of when evidence epistemically supports a conclusion. I think something like this might be on the right track. Evidence normically supports a claim when, roughly speaking, given that evidence, the claim would normally be true. This notion of normalcy does not reduce to statistical frequency. If the evidence obtains and yet p is false, some abnormality or malfunction has occurred. The error demands explanation. When evidence statistically indicates p but p is false, by contrast, it is not really an error; it is simply a case of ‘you win some you lose some’.

I mentioned earlier that I think philosophers must invoke ‘normalcy’ in theories. This is one place this idea shows up. Smith understands normalcy through a possible worlds framework. I am a little dubious about this way of understanding normalcy. But I think there is promise in thinking about evidential support by invoking the idea of what normally follows from the evidence. I think focusing on the nature of reasoning will illuminate legal standards of proof.

Blackstone has famously claimed that it is worse to convict a single innocent than letting ten guilty persons escape. If Blackstone’s claim is true, wouldn’t then legal systems have, in principle, good reasons to adopt more stringent standards of proof than ‘proof beyond reasonable doubt’ in order to minimise false convictions? What is your take on this? You’ve already briefly mentioned that you disagree with Larry Laudan’s proposal that we should lower our standard of proof for criminal conviction in some cases. But can you tell us a little more about your own views?

Georgi: Yes, Blackstone famously wrote ‘It is better that ten guilty persons escape than that one innocent suffer’. Some people use this to directly support the idea that the standard of proof for criminal conviction has to be over 90% confidence. Even leaving aside my worries about quantifying the standard, this is too quick. For the Blackstone ratio to support a particular threshold for a standard of proof we would also need to consider the prevalence of guilt and innocence among those tried, whether jurors are prone to high or low confidence in guilt, which kind of error (false convictions or false acquittal) is easier, and so forth. But, yes, Blackstone’s idea is used to support a stringent standard of proof.

Laudan argues that the standard of proof can adjust a trade-off between false acquittals and false convictions. He applies expected utility theory to argue that the standard of proof, at least for violent crime, should be a lot lower. He claims there is currently too great a risk of being attacked by people who were previously falsely acquitted of violent crime. Since this harm dwarfs the risk of false conviction for violent crime, we should lower the burden of proof to prevent the guilty from being acquitted.

This argument relies on social statistics. Laudan claims that ‘the lifetime risk an average American runs of being a victim of a serious, violent crime (excluding homicide) is about 83%’ and ‘each false acquittal enables on average more than 36 crimes, including seven violent ones’. Given the relevant risk and harms, Laudan suggests that the criminal standard of proof should be about 56 to 67%. He suggests the legal system should aim for a ratio of false acquittals to false convictions of about 2:1. This is starkly different from Blackstone’s oft-cited 10:1.

I dispute Laudan’s use of statistics. I argue that Laudan overestimates recidivism rates, crime rates, and the chances of being a victim. And he underestimates the harms of incarceration and wrongful conviction. He also overlooks the crimes that occur in prison or because of prison, and various other factors. (Prison crime is rarely represented in official crime statistics. The incarcerated are currently so excluded from society!)

So I argue that Laudan’s arguments are flawed. I defend the stringent standard of proof for criminal conviction. I don’t think the standard of proof should be determined by the contingencies of crime rates or by an attempt to maximise expected utility. But even if it should be, Laudan’s arguments fail.

In a forthcoming paper you defend evidentialism (roughly, the view that only factors relative to evidence and truth ground epistemic justification) against the moral encroachment objection. Could you briefly explain the objection and why do you think evidentialism still stands against it?

Georgi: Sure. Yes, so moral encroachment holds that moral features of a belief can affect the epistemic justification of that belief. So this view denies that the epistemic realm is ‘pure’, in the sense that it is only responsive to evidence (or other truth-relevant factors).

In Tamar Szabó Gendler’s 2011 paper ‘On the Epistemic Costs of Implicit Bias’ she introduces an incident that the historian John Hope Franklin describes in his autobiography Mirror to America. Franklin was hosting a gathering at his Washington D.C. social club, The Cosmos Club. As Franklin writes, ‘It was during our stroll through the club that a white woman called me out, presented me with her coat check, and ordered me to bring her coat. I patiently told her that if she would present her coat to a uniformed attendant, “and all of the club attendants were in uniform,” perhaps she could get her coat’.

Almost every attendant at the Cosmos Club is Black, and few members of the club are Black. This demographic distribution almost certainly led to the woman’s false belief that Franklin is an attendant. Many epistemologists argue that beliefs like this are well supported by the evidence, but are morally impermissible. But it looks like a problem if we can be epistemically required to believe claims that we morally ought not believe.

Moral encroachment argues that moral and epistemic demands aren’t in tension: In cases like the one Franklin describes, the moral features render the belief epistemically unjustified. There are various versions of moral encroachment, but perhaps the simplest one to explain is that the high moral stakes mean that more evidence is needed for the belief to be justified. (See Rima Basu’s ‘The Moral Stakes of Racist Beliefs’, for example.)

Basu’s argument for moral encroachment requires a person whose beliefs are epistemically impeccable—well supported by the evidence and conscientiously considered—yet morally wrong because, say, racist or sexist. My contention is that no such belief can exist: If a belief is morally wrong then there is some corresponding prior epistemic error. And usually there are other moral errors going on in the background, such as disparaging attitudes. It would take a long time to list common errors, and in the question you mention being brief, so perhaps I should just say that I describe a lot of these errors in ‘Evidentialism and Moral Encroachment’. I also argue that the moral evaluation of a belief depends on the broader understanding in which the belief is embedded. Some beliefs can be morally evaluated by themselves, such as absurdly racist ones that aren’t responsive to evidence at all. But for many beliefs, we can’t evaluate it morally without learning other things the person believes, and how their beliefs fit together.

If you take the above example, well, Franklin wasn’t wearing a uniform, and the staff members were. He was eighty years old. In general people make the mistake of focusing too much on race and gender, by implicitly assuming that they are highly probative features, and thereby overlooking other evidence. In general, there is a lot of error and bias. And this everyday, ubiquitous error can be explained by orthodox epistemology, without invoking moral encroachment.

In a way I think that positing moral encroachment (in response to incidents like the one Franklin describes) gives too much credit to racist beliefs. I just want to say that racist beliefs make old fashioned, orthodox-epistemology errors: they don’t reflect the available evidence. We don’t need to adapt our understanding of epistemic normativity. Racist beliefs aren’t epistemically justified; and this is simply because they are poorly formed.

That was a quick overview, but I think this topic raises a lot of interesting questions.

Do you have additional future projects in legal or moral philosophy?

Georgi: I would like to think more about ideas of reasonableness in the law. Reasonableness is invoked in the ‘beyond reasonable doubt’ standard of proof and the ‘reasonable person’ standard. Reasonable fears are required for self-defence, for example. I am interested in how moral and epistemic considerations influence what is reasonable.

I have also been wondering whether the adversarial structure of some legal systems itself plays a central role in satisfying ‘proof beyond reasonable doubt’, and I am developing a feminist challenge to using the preponderance of evidence standard of proof for Title IX investigations.

I have been writing the Oxford Bibliographies Online (OBO) entry for legal epistemology. This has given me a lot of ideas to pursue. It’s been an eye-opening exercise, seeing how many different topics fall under the ambit of legal epistemology. The law is indeed up to its neck in epistemology! I will post the OBO entry on my website, at least for the next week or two, even though it isn’t finished, so that readers can have a sneak peak. Feedback is very welcome. 

Is there any kind of work you would like to see published more often? 

Georgi: I think philosophy is extremely difficult. And so I would like to see more epistemic humility in the pursuit of philosophical understanding. I like the idea of using essays and presentations to explore ideas, investigate terrains, raise questions. A lot of essays are either arguments for p, or arguments against p. But it would be good to create more space for essays that say ‘This topic is interesting. Here are some thoughts, some questions, and here is where I am confused.’ I am not sure whether that counts as a ‘kind of work’ in the sense you mean.

I am happy to see the ‘applied turn’ in epistemology, and I am looking forward to reading more work in this area.

Most challenging part of grad school in philosophy? 

Georgi: I enjoyed graduate school. I developed some habits and approaches—‘study skills’, I suppose—during my Master’s degree that helped me to flourish during my doctorate. And I felt supported at Rutgers. It wasn’t really a challenge or ordeal, thankfully.

I think perhaps a significant challenge of graduate school is how the years of graduate school are dominated by the job market. It looms over everything, and can cast a shadow on the experience.

Could you tell us a bit more about the habits and approaches that helped you during your doctorate?

Georgi: I can try! None of this is particularly original. I will focus on two strategies that help me with projects such as essays. I think it boils down to four components. These are: (i) breaking larger tasks into smaller ones, (ii) creating the motivations and incentives for doing the small, manageable tasks, (iii) removing the blockages to starting, which for philosophy writing (for me) is something like fearing that the stakes are high, and so procrastinating rather than starting to write, and finally (iv) staying organised.

Three of these components are largely addressed (for me) by having an attractive to-do list. The list becomes more attractive as you complete the tasks, which provides some of the incentive. So, it could be neat handwriting on coloured card or pattered paper, for example, and rubber stamps instead of mere check marks. A ‘to-do’ list for an essay, for example, will start with a few of the readings you need to review. It will have basic essential tasks, such as finding out the word limit, or checking the requirements. It will then have about six entries for creating the plan. It is important that the entries are conceived of as small, manageable tasks, so a single entry for ‘write plan’ might seem too large and daunting, and so isn’t in the spirit of this suggestion. Then maybe every 400 words of writing is a new entry. And then there are entries for editing, and so on. It’s a simple thing, but it provides a unified way to break up tasks, create incentives, and stay organised. My lists also help in other ways, such as by providing a vivid reminder that I am approaching the essay’s word limit.

I mentioned the essay plan might be six different entries in the to-do list. My essay plans are extensive. They lay out the whole essay, almost down to the sentence level. Whenever I start a new essay I tape pieces of plain paper together to make larger sheets. During my Master’s thesis I taped six or eight pieces of paper together to make huge sheets! Now I just tape two or four. And I write the plan on those sheets. The plan usually takes about twelve large sheets in total. The larger size is helpful because it reduces the stakes: I can write a sentence, and then leave a couple of inches before the next one. If I need to squeeze in some text, or move it around, there is space. So writing the initial plan feels ‘low stakes’. It removes the blockage I mentioned above. The large sheets also help me see the ‘overview’—the ‘big picture’ perspective—on the essay. Once I have made the plan, typing up feels easy. Since the plan is already made, the writing simply flows and isn’t interrupted by, for example, wondering what part to write about next. Each paragraph is already determined, almost to the level of each sentence. It feels more like ‘typing up’ than writing. It only takes a day or two to ‘transfer’ the plan into an essay. The use of these large, extensive plans was a breakthrough experience during my Master’s degree.

I am just describing what worked for me, of course. So for some people this kind of approach will make sense, and for others it might sound rather odd.

When not doing philosophy… ? 

Georgi: I love community circus. I love to perform acrobalance. I spin fire, stilt, hula hoop, and I do some aerial acrobatics too. I played Peter Pan recently, and the scene where I taught the Darling children to fly was a bungee act. I was hanging from the ceiling on bungees, bouncing and flipping around. At the end of the act – by design! – I dropped head first towards the stage.

I think community circus is a special activity. By the term ‘community’ circus I mean something like ‘welcoming of various skill levels’ and ‘no performing animals’.

Recently I have started reading novels a lot more. It has been nice to remember how enjoyable reading ‘for pleasure’ is. In the last few years so much of my reading has been aiming at some other goal: to learn about society, to fill out my education more, to work on my philosophy research. But recently I have just been reading for fun. It has been wonderful.

Could you list two philosophy books and two non-philosophy books that have influenced you the most?

Georgi: I find conversations extremely fruitful. I think conversing with people has been central to my development. I have benefitted so much from my interactions with the work of Duncan Pritchard and Ernie Sosa—both reading their work and having conversations.

I am going to cheat and say three philosophy books: Catherine Elgin’s Considered Judgment, Jonathan Kvanvig’s The Value of Knowledge and the Pursuit of Understanding, and Edward Craig’s Knowledge and the State of Nature. These are all superb and shaped my thinking.

As for non-philosophy, in high school I read Laurie Lee’s As I Stepped Out One Midsummer Morning. It impressed on me the beauty and possibility of an itinerant, unusual life. It is so lyrical, poetic. Lee wanders around in Andalusia. After high school, my best friend and I went to Andalusia. We wandered around, trying to breathe Lee’s air. (On reflection, I think we recreated Bill Bryson more than Laurie Lee.)

The books that had the most influence were the ones I read when I was younger, because they shaped the later years. For philosophy, that means books I read as an undergraduate, and for non-philosophy that means things I read in high school. But a book I read recently that helped develop my understanding is Douglas Blackmon’s Slavery by Another Name. It’s about prisons and the legal system in the US. The subtitle is The Re-Enslavement of Black Americans from the Civil War to World War II. It was illuminating.

You can find more information about Georgi and most of her papers at her personal website.