On 14 October 2020, the High Court handed down judgment in ABT17 v Minister for Immigration and Border Protection. While the result was unanimous – and, at least in my view, clearly correct – the case produced four judgments, embodying two distinct strands of reasoning (one contained in the joint judgment of Kiefel CJ, Bell, Gageler and Keane JJ, and one in the separate judgments of Nettle J, Gordon J and Edelman J). Unfortunately, neither strand of reasoning is terribly satisfactory. In particular, while the reasoning contained in the joint judgment is far preferable to that contained in the separate judgments, it avoids addressing the difficult question whether appellate courts should simply acknowledge that human decision makers have no reliable capacity to gauge a person’s honesty from their ‘demeanour’, and that a fact-finding process that rejects a person’s evidence solely on the basis of demeanour should be regarded as fundamentally irrational. The minority judgments, on the other hand, are difficult to reconcile with the traditional understanding of ‘merits review’ in Australian administrative law, ascribe to the concept of ‘demeanour’ a probative value that it cannot rationally have, and seem to suffer from a fundamental internal contradiction (pursuant to which a person’s ‘demeanour’ is treated as having substantial probative value, but as nevertheless not constituting ‘information’). Unfortunately, the judgments in ABT17 are likely to encourage administrative decision makers to base their evaluations of evidence on intangible matters like ‘demeanour’ that lack probative value and to rely on assessments of demeanour as a mechanism for insulating logically infirm decisions from judicial review.
What does ‘demeanour’ mean?
To understand the two strands of reasoning in ABT17, it is necessary first to understand what ‘demeanour’ means in the context of legal fact finding. Broadly speaking, it refers to the collection of non-verbal inputs that make a conscious, subconscious or unconscious impression on a decision maker when a person answers a question or recounts a narrative. ‘Demeanour’ comprises things like tone of voice, eye and hand movement, hesitation, and word choice that have no rational connection with the likelihood that an account of events is or is not objectively true, but that a decision maker may take into account in deciding whether to believe what a person says. Demeanour is distinct from those matters that rationally go to a person’s credit (such as whether they have a motive to lie and whether they have a history of dishonesty).
We all like to believe that by paying careful attention to a person’s demeanour, we can tell whether they are telling the truth or attempting to deceive us. And we are all wrong. In fact, humans are terrible at spotting lies – our innate lie detectors are no more accurate than tarot readings or astrological charts. And people who believe they are good at spotting lies are no better at it than the rest of us. In fact, our belief that we can spot liars, combined with our unconscious tendency to believe people who resemble us in some way (and our tendency to treat physically attractive people more favourably than others – beauty is far from skin deep!), means that observing a person’s demeanour is just as likely to enable a liar to trick us (or to enable us to feel justified in disbelieving a truthful narrator who does not look like us, or who is physically unappealing to us) as it is to enable us to accurately detect a lie.
In appellate courts and law reform bodies in the Anglo-Australian world, there has been some recognition of the human animal’s total lack of ability to detect deception and of the implications this should have for legal proceedings. Thus in Pell v The Queen, Weinberg JA noted that:
The Australian Law Reform Commission, in its ground breaking work that led ultimately to the enactment of the Uniform Evidence Law, reviewed a great deal of psychological research concerning the demeanour of witnesses. That research almost universally concluded that facial reaction and bodily behaviour were unlikely to assist in arriving at a valid conclusion about the evidence of most witnesses.
Members of the High Court have also expressed doubts about the probative value of ‘demeanour’. Thus almost twenty years ago, Gleeson CJ, Gummow and Kirby JJ – who, like Weinberg JA in Pell, cited Lord Atkin’s statement in Société d’avances Commerciales (Société Anomyne Egyptienne) v Merchants’ Marine Insurance Co that ‘an ounce of intrinsic merit or demerit in the evidence … is worth pounds of demeanour’ – observed that ‘in recent years, judges have become more aware of scientific research that has cast doubt on the ability of judges (or anyone else) to tell truth from falsehood accurately on the basis of such appearances.’
The problem with such statements is that they don’t go far enough. They ascribe limited probative value to something that has no reliable probative value at all, and that often has a negative probative value. The extent to which a decision maker’s assessment of the ‘demeanour’ of a party or witness (or, for that matter, of a lawyer) may in fact undermine the rationality of legal decision making is not something our legal system has ever really grappled with. And in the context of judicial proceedings, this is understandable. Disputes inter partes often require judges and juries to choose between two or more conflicting accounts of events, each of which is inherently plausible, and none of which is contradicted by physical evidence. In these circumstances, confidence in the legal system is probably better preserved by appealing to a shared superstition like our belief in our capacity to read others’ demeanours than by, say, throwing a bunch of witness statements down a staircase and accepting as true whichever one travels the farthest. Unfortunately, a perfectly rational legal system is not possible, nor would perfect rationality even assist much. Disputes have to be resolved, and a ‘perfect’ rational evidentiary foundation for resolving them is rarely (if ever) available. Still, even in this context, it is important for judges and juries to be aware that their perceptions of witnesses’ truthfulness are just that – idiosyncratic impressions that are no more likely than the flip of a coin to correspond with objective reality. Fortunately, many judges have at least some degree of awareness of this. Further, in judicial proceedings, the proper application of burdens and standards of proof – which are not value-neutral adjectival rules, but which in fact have a substantive policy content of their own – may assist in deciding which of multiple competing narratives should be accepted.
But administrative decision making generally does not involve deciding between the competing claims of different parties, and rarely requires a decision maker to choose between multiple Rashomon-like accounts of the same event. Rather, administrative decision making ordinarily involves balancing a natural or legal person’s interest in obtaining some benefit (or avoiding some detriment) under a statutory provision against the diffuse public interest in ensuring that the provision is administered lawfully. The decision maker’s task is to determine whether, on the basis of the material placed before it by the person whose interests are liable to be affected by its decision, the relevant legal test is satisfied. In this context, it is difficult to see what purpose is served by basing findings of fact on matters that lack any rational probative value. It is especially difficult to see how ‘demeanour’ can be of assistance in a decision making context that – like the context in which protection visa applications are decided – allows the decision maker to decide that a past event may have occurred (and to base its decision on this possibility) and thus does not oblige the decision maker either to accept or reject every account of events placed before it.
The facts in ABT17
In ABT17, the appellant was a Tamil citizen of Sri Lanka. He applied for a protection visa on the grounds that if he was required to return to Sri Lanka, he would be persecuted because of his Tamil ethnicity and because he would be perceived by the Sri Lankan authorities as having supported the Liberation Tigers of Tamil Eelam (the LTTE) during the Sri Lankan Civil War. In support of this claim, the appellant alleged that before he left Sri Lanka, he had been detained on a number of occasions by the country’s security forces, who had targeted him ‘with a view to him confessing he was LTTE.’ During an interview with the Minister’s delegate, the appellant disclosed that on one occasion, he had been detained by Sri Lankan security forces for six days, during which he had been beaten and sexually assaulted (something that, according to a number of human rights groups, remains common in Sri Lankan detention facilities). The delegate found that these allegations were plausible, but considered that country information showed that young Tamil men were not, as a class, likely to be persecuted by the Sri Lankan government because of their perceived links to the LTTE. The delegate therefore rejected the appellant’s application for a protection visa.
The delegate’s decision was referred to the Immigration Assessment Authority (the IAA) for review under Pt 7AA of the Migration Act 1958 (Cth). Pursuant to Pt 7AA, the IAA must review a decision referred to it and must either affirm the decision or remit it to the Minister (or his delegate) for reconsideration. Importantly, Pt 7AA contemplates that review by the IAA will generally be done ‘on the papers’. This is subject to s 473DC, which provides that the IAA may interview an applicant for the purpose of getting ‘new information’ (being information that was not before the Minister or his delegate when the decision under review was made and that the IAA considers may be relevant to the applicant’s case).
In ABT17, the IAA did not invite the applicant to be interviewed, and instead based its review on documentary material and on an audio – not audio-visual – recording of the applicant’s interview with the Minister’s delegate. Based on these matters, the IAA considered that the applicant ‘at times sounded vague and hesitant’ and concluded that he had ‘exaggerated and embellished some of the incidents he relies on in order to enhance his profile and claims for protection.’ The IAA went on to state that country information showed that conditions on the ground in Sri Lanka had changed since the end of the country’s civil war, and that young Tamil men were not generally at risk of persecution. It therefore concluded that ‘[t]aking into consideration the number of years that have elapsed since he left, his personal circumstances and the country information referred to’, there was not a real risk that the appellant would be persecuted (or would suffer ‘significant harm’) if he was required to return to Sri Lanka.
It is important to note here what the IAA’s fact-finding process entailed. In short, the IAA rejected as an outright lie the appellant’s claim that he had been sexually assaulted in a detention facility in a country where, according to credible information, sexual assault is routinely practised in detention facilities. And it did this on the basis of its perception that the applicant’s account of events was, in some entirely undefined way, ‘vague’ and ‘hesitant’. The fact that this type of fact-finding process will be familiar to anyone who has ever read a migration-related administrative decision should not distract from how fundamentally absurd it is. It is simply not rational for a decision maker to wholly reject a fundamentally plausible narrative on the basis of nothing more probative than the fact that the narrator… just kind of sounds like a liar, or something? And dressing the reasoning up with legal-sounding language about vagueness and hesitancy (or ‘demeanour’) does not make it any less irrational.
Proceedings below the High Court
The appellant sought judicial review of the IAA’s decision in the Federal Circuit Court, where he was refused leave to amend his originating process to add an argument to the effect that the IAA had acted unreasonably by failing to invite him to an interview for the purpose of assessing his demeanour in person. On appeal to the Federal Court, Bromberg J considered that it was reasonably arguable that the IAA had acted unreasonably by failing to interview the appellant. However, his Honour held that the IAA’s assessment of the country information concerning conditions in Sri Lanka had provided an independent basis for its decision, and that any error in its assessment of the appellant’s credibility and personal history therefore had not been ‘material’ (in the sense that the decision could not realistically have been different if the IAA had not erred).
The majority judgment in the High Court
The majority referred to Lord Atkin’s dictum in Société d’avances Commerciales and, at least to some degree, acknowledged the limited value of demeanour as a means of assessing the truth of a person’s account of events. Nevertheless, it considered that the fact that the IAA was possessed only of an audio recording of the appellant’s interview with the Minister’s delegate created an ‘informational gap’, such that the IAA lacked ‘a visual impression of how the referred applicant appeared during the interview – his or her demeanour.’ According to the majority, it was open to the IAA to bridge the ‘informational gap’ by conducting an interview with the appellant under s 473DC of the Migration Act. This was because:
The Authority’s own visual impression of the referred applicant’s appearance during such an interview would necessarily constitute new information within the power of the Authority to get because it would communicate knowledge of an evidentiary nature which would be open to be considered by the Authority to have the potential to bear on the Authority’s assessment of the referred applicant’s credibility and which was not before the Minister when the delegate made the referred decision.
In the circumstances of the appellant’s case, the procedural aspect of ‘legal reasonableness’ had required the IAA to interview the appellant before rejecting his account of events. This was because:
[T]he Authority will act unreasonably if, without good reason, it does not invite a referred applicant to an interview in order to gauge his or her demeanour for itself before it decides to reject an account given by the referred applicant in an audio recorded interview which the delegate accepted in making the referred decision wholly or substantially on the basis of its own assessment of the manner in which that account was given. That is what happened in this case.
According to the majority, the question of materiality did not arise, because the IAA had explicitly based its decision on a combination of its interpretation of the relevant country information and its assessment of the appellant’s credibility. In these circumstances, it could not be said that the country information provided an ‘independent basis’ for the decision.
The minority judgments
Justice Nettle proceeded on the basis that the process contemplated by Pt 7AA of the Migration Act – a statutory process that does not (and could not) involve the exercise of judicial power, that does not involve choosing between competing ‘cases’, and that does not require all relevant allegations of past fact to be accepted or rejected in a definitive way – is ‘analogous to a process of appeal by way of rehearing from the judgment of a judge sitting alone.’ Based on this, his Honour considered that the IAA should defer to any assessments of credit contained in a decision under review, and should not depart from the findings contained in a decision unless ‘the delegate’s findings are contrary to incontrovertible facts or uncontested testimony, or glaringly improbable, or contrary to compelling inferences, or otherwise erroneous’. For these reasons, Nettle J considered that the vice in the IAA’s decision was not that it had rejected the appellant’s account of events on the basis of nothing more probative than its subjective assessment of his demeanour (or that it had failed to place itself in a position where it could fully appreciate both the auditory and visual aspects of the appellant’s demeanour), but that the IAA had not been sufficiently deferent to the delegate’s subjective assessment of the appellant’s demeanour – that is, ‘[g]iven that the delegate’s acceptance of the appellant’s claims was thus to a significant extent informed by the delegate’s assessment of the appellant’s demeanour as derived from the benefit of seeing and hearing the appellant … it was legally unreasonable for the IAA to depart from the delegate’s assessment of the plausibility of the appellant’s claims.’ But despite concluding that ‘demeanour’ is of such probative value that a primary decision maker’s assessment of it should be deferred by a merits reviewer, his Honour (without deciding the issue) expressed substantial doubt as to whether it was ‘information’ for the purposes of s 473DC.
Much of Gordon J’s judgment was devoted to the principles governing the doctrine of materiality (on which Gordon J’s views are, in my view, in many ways preferable to those of the majority of the Court). Given that the majority did not consider materiality to be relevant to the appellant’s case, this aspect of her Honour’s judgment is perhaps best viewed as a ‘dress rehearsal’ for any future case in which the Court may decide to reconsider the correctness of the majority’s judgment in Minister for Immigration and Border Protection v SZMTA. On the other issues raised by the appellant’s case, Gordon J expressed broad agreement with Nettle J, stating that ‘[i]n order to form and act on its own assessment of the relevance of the review material to its review, the Authority may not, without sufficient reason, reject that part or those parts of the review material based on demeanour.’ However, on the question whether s 473DC authorised the IAA to interview an appellant for the purpose of independently assessing their demeanour, Gordon J was more emphatic than Nettle J, stating that ‘having the appellant repeat his account of what had happened to him in an interview with the Authority is not to get new information.’
Justice Edelman broadly agreed with Nettle J and Gordon J, holding that ‘it was a legal error for the Authority, which did not have the opportunity to assess the demeanour of the appellant including the demonstration of his scarring, to depart from the delegate’s assessment of the appellant’s credibility.’ Like Gordon J, his Honour considered that s 473DC did not authorise the IAA to reassess an applicant’s demeanour, stating that an applicant ‘does not “give” their demeanour. Rather, the demeanour of, or manner in which the evidence is given by, a referred applicant is a matter “on which the value of [the] evidence depends”.’
The result in ABT17 was plainly correct. It is simply absurd for a decision maker to reject as fiction – not to find that it might not be true, but to wholly reject it as a lie – a credible narrative on the basis that an audio recording of that narrative makes the narrator sound like a liar in some undefined way. In this regard, even if one accepts that ‘demeanour’ has some probative value that science cannot measure, that probative value must surely derive from some combination of auditory and visual stimuli. The fact finding process employed by the IAA was thus fundamentally deficient, and its decision was rightly quashed. But the judgments in ABT17 are, in my view, deeply concerning, because they appeal to a faith in the probative value of ‘demeanour’ that cannot be sustained by empirical data. On the majority’s view, provided that a decision maker goes through the motions of observing a person’s ‘demeanour’ in both its auditory and visual dimensions, it is open to the decision maker to reject as fiction a plausible account of events that is consistent with objective evidence on the basis of nothing more than a hunch that the person is not telling the truth (or, which is far worse, a conscious or unconscious predisposition to believe that people like that person generally do not tell the truth). The majority’s judgment is thus likely to encourage decision makers to shield their decisions from judicial review by couching them in terms of unexaminable observations about ‘demeanour’ – something that generally has no probative value at all. That is, the majority judgment, though it reaches the correct result, functions as a blueprint for insulating irrational decisions from judicial review.
But it seems to me that the minority judgments are even more problematic. In particular, when one considers that a visa application will only reach the IAA if it has been rejected by the Minister’s delegate, and that the delegate’s decision is thus far more likely to be based on a negative than a positive assessment of the applicant’s demeanour, the minority’s position would in many cases require the IAA to defer to negative findings on questions of credit that lacked any real evidentiary basis. This approach would deny many applicants the true ‘review’ that they are entitled to under Pt 7AA. Further, at a basic level of logic, it is very difficult to reconcile the minority’s view that ‘demeanour’ has sufficient probative value that a primary decision maker’s findings based on it should be deferred to with the conclusion that it does not constitute ‘information’. While I personally agree that a person’s demeanour is not information, this is because I firmly believe it generally has no probative value. I do not understand how one can logically both believe that demeanour has probative value and deny that it constitutes information.
Finally, ABT17 suggests that in cases where legal ‘irrationality’ (as opposed to ‘unreasonableness’) is explicitly relied on as a ground of review (in a way that the procedural history of ABT17 prevented), the High Court is unlikely to take the strict logician’s approach to jurisdictional error in fact finding that the Federal Court has taken in some recent cases. Alternatively, it may simply be that ABT17 was a case in which it was patently obvious that the impugned decision was tainted by legal error, and in which the procedural history of the matter – in which ‘irrationality’ was not argued as a ground of review below and could not have been raised for the first time in the High Court – resulted in the various members of the Court putting forward a series of less than perfect justifications for setting aside the decision. Either way, an unfortunate by-product of ABT17 is likely to be more administrative decisions in which difficult evidentiary questions are simply waved away with vague incantations about such non-probative matters as ‘demeanour’, ‘tone’, ‘appearance’, and ‘hesitancy’.