The ever-expanding boundaries of the irrationality ground of judicial review (quarantine length post)

About two weeks (or seven binge-watched seasons of 30 Rock, or an amount of time spent playing Assassin’s Creed that I’d rather not calculate and am not particularly proud of) ago, a Full Court of the Federal Court handed down judgment in BFH16 v Minister for Home Affairs, holding (by a 2-1 majority) that the Administrative Appeals Tribunal had fallen into jurisdictional error by failing to assess the appellants’ protection visa claims in a manner that was legally rational. The judgment is notable for two things: first, the extent to which the majority was willing to interrogate the implicit assumptions and generalisations that underpinned the Tribunal’s findings on the key question of fact raised by the appellants’ claims; and second, the Court’s refusal to draw a bright line between the ‘irrationality’ and ‘unreasonableness’ grounds of judicial review. On the first point, the majority’s judgment illustrates the extent to which the Federal Court continues (in my view, correctly) to expand the boundaries of judicial review for legal irrationality and to do away with any supposed distinction between the ‘irrationality’ and ‘no evidence’ grounds of review. With respect to the second point, it illustrates the Court’s (in my view, somewhat unfortunate) tendency to eschew the formulation of ‘categories’ and ‘taxonomies’ of jurisdictional error at all and to approach the task of judicial review in a manner that does not always promote predictability or consistency in judicial decision making.

The appellants’ claims and the Tribunal’s decision

The two male appellants (‘BFH16’ and ‘BFI16′) were from Pakistan and claimed to have commenced a long-term sexual and romantic relationship in Australia after they began living together as part of a share house arrangement in 2010. Each appellant claimed that this was the first same sex relationship he had been in. The appellants claimed to be entitled to protection visas on the grounds that if they were required to return to Pakistan, there was a real risk that they would face persecution because of their sexuality. The Tribunal does not seem to have had any doubt that gay men face a real risk of persecution in Pakistan. However, it found that the applicants had never been in a relationship of the kind they claimed, and that they had instead concocted their story in order to advance their visa claims.

Before concluding that the appellants’ relationship was not genuine, the Tribunal put to each appellant a series of questions about what he claimed had happened in the days following their first sexual experience together. In response to these questions, BFI16 said that the appellants did not talk about the experience the next day (instead they ‘just went about their normal routine’) and that they did not have sex again for a couple of weeks. BFH16, on the other hand, said that the appellants talked openly with each other that night (and the next day) and had sex again a couple of days later. The Tribunal also noted a number of other matters that it considered were inconsistent with the appellants’ being in a genuine relationship. Indeed, as the appellants submitted before the Full Court, there were five key reasons why the Tribunal concluded that the appellants had never been in a sexual or romantic relationship. However, to understand the majority’s judgment, it is only necessary to set out two aspects of the Tribunal’s reasoning:

1. First, the Tribunal considered that after the appellants first had sex, ‘there would be much that they would want to discuss with each other’. It therefore dismissed as ‘totally implausible [BFI16’s] comments that the next day they just went about their normal routine and that he cannot remember when they discussed the implications of what had happened.’
2. Second, the Tribunal did not accept that the appellants ‘would not be able to recall two very significant details of what happened in the aftermath of what both claimed was their first experience of sexual intercourse and their first significant gay sexual experience, that is, when they discussed the implications for them individually and as a couple of what had happened; and when they next had sexual intercourse.’

Accordingly, the Tribunal found that the appellants had attempted to ‘create a “gay profile” for the purposes of supporting their claims to protection, but [did] not consider that this [was] a genuine profile.’ It therefore found that there was no real risk that the appellants would be persecuted on the basis of their sexuality if they were required to return to Pakistan, and that they did not meet the criteria for the grant of a protection visa.

The judgment of the Full Court

The majority (Murphy and O’Bryan JJ) held that ‘the Tribunal’s reasoning with respect to the immediate aftermath of the appellants’ first sexual experience [was] logically flawed and thereby irrational.’ In this regard, the majority considered that the Tribunal’s conclusions depended on ‘assumptions about the expected psychological response of the appellants, and particularly BFH16, to their first sexual encounter.’ Their Honours went on to describe these assumptions:

The assumptions are twofold: that the appellants would not go about their normal routine but would immediately in some manner discuss the implications of their first sexual encounter; and that the appellants would remember doing so and would also remember (with a degree of precision reflecting the difference between 2 to 3 days or 2 weeks) when they next had sex. The Tribunal implicitly reasoned that the fact that BFH16 in particular did not have the expected psychological response is probative of (makes more likely) the conclusion that the appellants have fabricated their claimed first sexual encounter.

The majority considered that these assumptions were not rational, in that they were not grounded in evidence placed before the Tribunal or in ordinary human experience:

It cannot be said that the psychological reactions of a couple to their first sexual encounter are matters of common human experience. Indeed, to the extent that anything can be said about such matters from common human experience, it would be that the psychological reactions of a couple to their first sexual encounter are likely to vary widely, reflecting the wide range of human emotional attributes. The assumptions made by the Tribunal about the expected psychological response of the appellants, and particularly BFH16, to their first sexual encounter could not be established without other evidence, perhaps psychological evidence. No other relevant evidence was before the Tribunal and, accordingly, the assumptions were not proved.

Their Honours considered that these flaws in the Tribunal’s reasoning process were ‘material’ to the Tribunal’s decision, in that the outcome of the appellants’ cases could realistically have been different if the Tribunal had not fallen into the logical error they identified.

Justice Snaden dissented, holding that ‘[t]here was no want of logic, rationalism or intelligible justification, nor any caprice, plain injustice or manifest unreasonableness, inherent in the Tribunal’s apparent expectation that the appellants would have discussed the implications of—and would be able to recall, consistently and with at least a degree of particularity, details related to—an event that would have been of such obvious significance to them.’

Analysing the majority’s analysis

The chain of reasoning employed by the Tribunal (and rejected by Murphy and O’Bryan JJ as irrational) can be summarised in the following way:

1. The appellants did not clearly recall the immediate aftermath of (what they claimed was) their first sexual experience together.
2. Men usually have a clear recollection of the immediate aftermath of their first sexual experience with a male partner.
3. Therefore, the appellants had never had a sexual experience together.

Applying the terminology formulated by the English philosopher Stephen Toulmin (and adopted by the American jurist John Henry Wigmore), Step 1 in this chain of reasoning can be described as ‘data’, Step 2 as a ‘warrant’, and Step 3 as an ‘inference’. Alternatively, Steps 1 and 2 can be described as the minor and major premises of a syllogism (the syllogism here being something akin to ‘Socrates doesn’t clearly remember the first time he had sex with a man > all men clearly remember the first time they had sex with a man > Socrates has never had sex with a man’). It was at Step 2 — the Tribunal’s use of an implicit assumption about the way an ordinary man would respond to his first sexual experience with another man as a warrant from which it could be inferred that the appellants’ relationship was not genuine — that the majority held the Tribunal had acted irrationality. According to their Honours, this warrant could not be grounded in ordinary human experience. Rather, it was the kind of warrant that could only be justified by empirical data drawn from some form of systematic inquiry (such as, perhaps, a questionnaire administered by a psychologist to a representative sample of men).

How does this fit in with the Federal Court’s irrationality jurisprudence?

The majority’s judgment illustrates the extent to which the Federal Court has expanded the reach of the irrationality ground of judicial review over the past few years. Some older judgments of the Court proceed on the basis that the irrationality ground of review is only available where the Tribunal’s ultimate decision that an applicant does not satisfy the criteria for the grant of a visa (its decision with respect to the ultimate question of ‘jurisdictional fact’ that the Tribunal is required to decide pursuant to s 65 of the Migration Act 1958 (Cth)) cannot be justified by reference to the evidence placed before it. This narrow approach is perhaps best exemplified by the judgment of Rares J in SZOOR v Minister for Immigration and Citizenship, in which his Honour considered that he was constrained by High Court authority to hold that ‘even if the decision-maker’s articulation of how and why he or she went from the facts to the decision is not rational or logical, if someone else could have done so on the evidence, the decision is not one that will be set aside.’ More recently, however, the Court has repeatedly endorsed the view that the entire process of ‘fact finding which leads to the end result’ of an administrative decision is amenable to review on the grounds of irrationality, and that an administrative decision may be quashed if instead of being based on findings supported by evidence, it is grounded in ‘speculation or guesswork, or (worse) assumptions based on material incapable of supporting those assumptions.’ This will be so even if the Tribunal’s ultimate decision was open to it on the totality of the evidence.

The majority’s judgment takes this reasoning a step further, holding that it is not just express findings of fact that are amenable to judicial review for irrationality. Rather, legal irrationality may be identified through a process of making explicit the implicit generalisations and assumptions relied on by a decision maker in drawing an inference or reaching a conclusion. This is a logical extension of the Court’s recent jurisprudence. However, it is also an approach that opens up the entirety of a decision maker’s reasoning process to judicial review. As Rares J’s judgment in SZOOR demonstrates, it is not entirely clear that this approach is supported by High Court authority (or that that court will not in some future case decide that the Federal Court has spent the last half a decade or so formulating and refining a fundamentally wrong approach to the question of legal irrationality).

The Federal Court’s approach is right

For my own part, I find the argument for confining the irrationality ground of review to cases where a decision maker’s final conclusion on a question of ‘jurisdictional fact’ cannot be justified on the totality of the evidence to be unpersuasive. This is so for a number of reasons:

1. While the term ‘jurisdictional fact’ has become an unavoidable part of the administrative law lexicon, it has arguably outlived its usefulness. In the way it has come to be used by the High Court, the expression ‘jurisdictional fact’ encompasses both ‘subjective’ and ‘objective’ jurisdictional facts. But these are two very different things. A subjective jurisdictional fact is a state of satisfaction as to the existence of a fact or (more commonly) a conclusion of mixed fact and law that a decision maker must reach before it can lawfully exercise a power. For example, under s 65 of the Migration Act, the relevant decision maker must grant a visa if it is satisfied that the applicant meets the legal criteria for the grant of that visa. The grant of a visa does not become invalid if a court finds that the applicant did not, as an objective fact, satisfy those criteria. An objective jurisdictional fact, on the other hand, is a state that must in fact exist (and, on an application for judicial review, be found by the reviewing court to exist) if an exercise of statutory power is to be valid. Using the same expression to describe two such different concepts is confusing and serves no purpose. Further, in the same way that (according to Voltaire, at least) the Holy Roman Empire was neither holy, nor Roman, nor an empire, jurisdictional facts are generally neither ‘facts’ nor ‘jurisdictional’ in nature. As noted above, the express preconditions prescribed by statute for the valid exercise of a mandatory or discretionary power more often take the form of conclusions of mixed fact and law than of purely factual findings; the decision maker’s task requires it both to decide whether the prescribed conclusion of mixed fact and law is established and to make findings with respect to any questions of pure fact that arise on the path to its conclusion. Further, it is not generally correct to say that a decision maker cannot exercise ‘jurisdiction’ unless a jurisdictional fact is established — the task of inquiring into and determining whether the jurisdictional fact is established is itself an exercise of jurisdiction, albeit of a ‘preliminary’ kind. For these reasons, it might (as Basten JA has suggested) be better for Australian administrative law to jettison the expression ‘jurisdictional fact’ and simply ‘describe the properly formed state of satisfaction as a precondition to a valid exercise of the power.’
2. It follows from the proposition that so-called ‘jurisdictional facts’ are rarely ‘pure’ facts that in conferring power upon a decision maker to reach a conclusion with respect to a question of jurisdictional fact, the legislature also confers upon the decision maker the power (and, to at least some extent, the duty) to make any findings of ‘pure’ fact that are necessary for it to reach that conclusion. Surely the legislature must be assumed to intend that a power conferred by it on a decision maker to make pure findings of fact will, no less than the ultimate decision-making power to which those findings relate, be exercised rationally and on the basis of probative evidence.
3. Finally, as the authors of the administrative law bible have pointed out, an approach to irrationality that looks only to whether the decision maker’s conclusion on a question of jurisdictional fact could have been reached on a rational view of the whole of the evidence positively invites a court engaged in judicial review to review the merits of an administrative decision by re-evaluating each and every piece of evidence and question of primary fact that it (not the decision maker) considers relevant to the decision (rather than maintaining its proper role of determining whether the decision maker has fallen into some material error in discharging its statutory task).

It therefore seems to me that the majority’s approach to the issue of irrationality in BFH16 is simply a logical extension of a fundamentally correct approach to that issue.

Doubts about the result

Even though the majority’s judgment seems to me to represent a fundamentally correct approach to the irrationality ground of review, I can understand why one might (like Snaden J) find it difficult to accept that the Tribunal’s reasoning process was truly ‘irrational’. In particular, it might be argued that in holding that the Tribunal could not lawfully establish the ‘warrant’ or ‘major premise’ of its reasoning without relying on expert evidence, the majority set the bar of rationality at a level that would generally require decision makers to refrain from making generalisations about human behaviour that were not supported by expert evidence. But it is difficult to see how such an approach to decision making could be put into practice. In this regard, while it is true that a warrant or major premise can (and, where possible, should) should be drawn from empirical evidence or systematic inquiry, the warrants ordinary people (including the people who sit on juries in criminal cases) use in everyday reasoning about human behaviour are more often the product of unstated assumptions and generalisations than they are of evidence or scientific data. The point is well illustrated in Andrew Palmer’s article ‘Why and How to Teach Proof’:

In my classes I start by asking students to analyse something fairly simple, such as a photograph. For example, I show them a photograph that appears to have been taken in the United States in the 1950s and which depicts a white baby being held by a black woman. I ask them to put forward an argument about the relationship between the two figures. Of course most students would almost immediately infer that the woman was the child’s nanny, but they would do so without being fully aware of the reasoning process that had led them to reach this conclusion. By having to articulate arguments for their conclusion, however, students are forced to become aware of the ingredients of evidential arguments. At the microcosmic level, these are essentially data or evidence (the baby looks relaxed and comfortable with the  woman), which combine with generalisations (babies are usually only relaxed and comfortable with adults with whom they are familiar) to provide the basis for inferences or conclusions (the baby is familiar with the woman).

In this example, the warrant from which the students draw their conclusion comes from a basic understanding of the way children relate to adults and, perhaps, a superficial knowledge of race relations in the United States during the post-war period, and not from any specific expertise in childcare or American history.

However, it is not always irrational to base a conclusion on an explicit or unstated assumption that lacks even the kind of limited evidentiary basis described in Palmer’s example. Many of the generalisations and assumptions that inform everyday reasoning relate to matters that are inherently difficult to analyse through the systematic collection of data or that have been the subject of inconclusive or contradictory studies. To take a topical example, attempts to study the effectiveness of face masks in preventing disease transmission outside of hospital settings have generally proved inconclusive, to the point where it is not really possible to provide an evidence-based recommendation that ordinary citizens should or should not wear face masks during everyday outdoor activities in order to prevent the spread of the SARS-Cov-2 virus. But this does not mean a person who wears (or recommends that others wear) a face mask is acting irrationally. A person with a basic intuition about how respiratory viruses spread can legitimately reason that placing a physical barrier in front of their nose and mouth is at least reasonably likely to lessen their chance of spreading a disease like COVID-19. This reasoning might be unsupported by empirical evidence (and may prove to be wrong), but it is not irrational. Faced with imperfect data, citizens and public health authorities have to make decisions, and may have to do so on the basis of their own limited experience and knowledge. Similarly, a decision maker trying to decide how an ordinary person would react to a particular situation will often have to do so in circumstances where there is simply no useful empirical data that it can draw on.

The majority’s decision raises the question whether any scientific data of the kind it thought might have informed the Tribunal’s reasoning even exists. If such data does not exist, it seems somewhat problematic to hold that the Tribunal acted irrationally in failing to find it and act consistently with it. Moreover, the majority’s approach could perhaps be seen as tantamount to reversing the onus of proof in judicial review proceedings by requiring a decision maker (or the contradictor defending the impugned decision) to affirmatively establish that each step in its reasoning was supported by empirical evidence. If one ignores the specific context of the appellants’ claims, it is arguable that the reasoning dismissed by Murphy and O’Bryan JJ as irrational consisted of nothing more than the proposition that people ordinarily remember in some detail what happened immediately after significant events in their lives. This may or may not be true — many (if not most) of the things people believe about the reliability of human memory are contradicted by experimental data. But the belief that decision makers (including juries) can make rational generalisations about what events a person is likely to remember accurately, and can use those generalisations to draw inferences about a person’s honesty, is fairly well entrenched in the Anglo-Australian system of justice. And appeal courts still (albeit to a lesser degree than they used to) tend to treat a trial judge or jury that has had the opportunity to observe the demeanour of a witness as having an advantage in assessing the witness’ truthfulness, even though (as Weinberg JA noted in his dissenting judgment in Pell v The Queen, a judgment I have only read in passing and with which I neither agree nor disagree) there is a substantial body of scientific evidence to show that observing a witness’ demeanour does not increase the accuracy with which a judge or jury is able to gauge the truth of the witness’ evidence. The nagging doubt I have about the majority’s judgment in BFH16 is that if one accepts that the Tribunal’s reasoning process was irrational, it seems to follow that most assessments of credibility made in legal proceedings are irrational. While this may well be the case, it is not clear to me that there is any practicable way for the legal system to put the assessment of witness’ credit on a sounder rational footing — unfortunately, the assessment of credit seems to me to be something that necessarily requires decision makers to rely on implicit and explicit generalisations about human behaviour, most of which are unlikely to have any scientific basis and many of which are likely to be completely wrong.

However, I think the preferable interpretation of Murphy and O’Bryan JJ’s reasons for judgment is that, at least in their Honours’ view, there was no acceptable non-scientific basis — no basis in ordinary human experience, common intuition, or folk wisdom — from which the Tribunal could generalise about the way a man might be expected to act in the aftermath of his first sexual experience with another man. In these circumstances, legal rationality required the Tribunal either to point to an empirical basis for such a generalisation or it to refrain from making any generalisation at all. Ultimately, in the context of judicial review proceedings, deciding whether or not something in fact falls within the scope of ordinary human experience is a task for the reviewing court, and it is certainly not clear to me that Murphy and O’Bryan JJ were wrong to hold that the Tribunal had taken official notice of a ‘fact’ that ordinary life experience is not capable of establishing.

Conflation of irrationality and unreasonableness grounds of review

In its judgment, the majority drew no distinction between the ‘irrationality’ and ‘no evidence’ grounds of review. This is consistent with other recent jurisprudence in the Federal Court, which has emphasised that the irrationality ground of review is concerned with findings of fact that have no basis in ordinary human experience or in probative evidence, and not just with patently outlandish assertions (‘His hair is thick, like a Bolshevik commissar. He made me watch a giraffe with the legs of a man!‘). The merging of these two grounds of review is a logical development. For the reasons discussed above, not every finding of fact that lacks support in empirical evidence is irrational, and administrative decision makers have to be able to have regard to their own experiences and intuitions — things that are not ‘evidence’ in the usual sense of the word — in making findings of fact. That is, they have to be able to take ‘official notice’ (the administrative law equivalent of ‘judicial notice’) of matters that are not the subject of evidence. Doing this will only lead to jurisdictional error if the decision maker takes official notice of a matter that is simply not one that ordinary human experience can rationally shed any light on.

However, the majority also declined to draw a clear distinction between the ‘irrationality’ and ‘unreasonableness’ grounds of review, stating that ‘[t]he distinction … has not been widely embraced and there is a considerable weight of authority against the adoption of rigid categories and formulae in the explication of principles of jurisdictional error.’ In my view, this was a mistake. As Derrington J  wrote in a recent judgment, it is preferable to ‘distinguish between alleged unreasonableness in the exercise of discretion and illogical or irrational reasoning in the course of reaching a state of satisfaction’ on a question of fact. In reviewing a decision with respect to the exercise of a discretionary power, a court must ask itself whether the decision maker’s exercise of discretion falls within the ‘area of decisional freedom‘ granted to it by the relevant statutory scheme. This is conceptually different from asking whether a finding of fact was made without any rational basis. Using different expressions — ‘unreasonableness’ and ‘irrationality’ — to describe these two processes ensures that the distinction is not overlooked. More broadly, The Federal Court’s tendency to avoid speaking in terms of ‘categories’ of jurisdictional error risks moving the process of judicial review away from a (relatively) predictable, rule-based system and into the realm of an ad hoc ‘I know it when I see it‘ jurisprudence, in which past decisions fail to give sufficient guidance to enable consistent resolution of future cases. At least in my view, a ‘taxonomical’ approach to jurisdictional error is more likely to promote consistency and predictability than one that wholly eschews categories and formulae.