What’s the problem with a ‘scientific hearing’ on animal tests?
For many years, animal rights protesters have called for a debate on the effectiveness of animal models. What could possibly be wrong with that? Debates are a normal part of our democracy, after all. Why not let the people hear the arguments and decide for themselves?
In short, because the debate the activists are demanding is rigged. Here’s how.
Debates are designed to make a case, like a lawyer, leaving in talking points that support a certain worldview and leaving out bits that don’t. They’re a form of storytelling, which contrasts with the job of science, which is to find out what’s true, warts and all.
Legal hearings are similar. Oral arguments in support of a motion are presented in order to decide things like whether a piece of evidence is admissible in a court case.
Whatever the outcome of a debate or hearing, we need science to inform moral choices and government policy, particularly since the government has an obligation to protect the interests of its citizens. We need to know what’s true, not what sounds attractive. Modern science as pioneered by the physiologists of the 1800s like the ‘father of physiology’ and ‘great priest of atheism’ Claude Bernard rejected religious, spiritual and other comforting biases in favour of that which could be experimentally demonstrated. Their whole thing was to create a better way of establishing what was real.
So, although it’s important from a social perspective for scientists to debate the use of animals in research, what if a scientist isn’t any good at debating? Should the facts they’re presenting be disregarded? Research shows that what can make a person convincing in a debate setting is their look, tone and demeanour, not necessarily the quality of their information, when the latter is of course the basis of a scientist’s expertise. Debate is an attempt to return to the loose system of hunches and kangaroo courts that existed prior to science.
Science results in data that can be interrogated in a structured and accountable way. Its findings can guide our thinking to more ethical and effective outcomes and better policy. That said, its conclusions can be revisited at any time, whereas a debate has an arbitrary timeline for responses.
We have a system for publishing information that, while imperfect, at least features peer review, very often open data and independently testable hypotheses. It also does not impede investigations into areas of interest (like the use of animals), including by people who are ideologically critical of animal use, and their evidence can be considered to see if and where it has value.
Indeed, those opposed to the use of animals in research regularly publish papers within this system. The PETA science consortium has published hundreds of papers. Nobody’s being silenced, although they shouldn’t be surprised if shoddy or inaccurate work is treated as such.
‘Scientific debate’
More recently, activists have been asking for a ‘scientific debate’, or ‘scientific hearing’, both of which are intended to undermine the slow march of science while trying to steal its clothes. It appears that this is a Tibetan-Buddhist form of debate with the arguments summed up – accurately or not – by the ‘champion’ of a debating position, but with a predetermined focus, set of definitions and list of admissible evidence. It reduces scientific findings to talking points which, if the campaigners’ previous work is any guide, will be misrepresented.
Worse still, the proposed debate focuses on a question of human/animal ‘prediction’ that has been addressed repeatedly in the past, most recently in 2017, where it was answered in a way that was able to draw on previously unpublished data to give the most up-to-date assessment possible. It found a human animal concordance of 86% on average, falsifying the idea that animals cannot predict human safety.
The situation is reminiscent of Creationists pushing ‘Intelligent Design’ a few years ago. They, too, had no data and an ideologue’s narrative so they wanted to ‘teach the controversy’ over evolution where no controversy exists.
The ‘hearing’ demanded by activists is similarly anti-scientific. It would appoint judges to pronounce on the nature of reality after a time-limited, evidence-limited, show-and-tell. The whole scheme has been sketched out several times by the activists who are pushing it in various similar forms, most recently here.
Cleverly, they initially use the reasoning behind, and language of, science to make the case for animal experiments undergoing scrutiny, which is all fine. However, they then add a twist – a conspiracy theory that the current system is corrupt and therefore a scientific question must be settled by a form of debate.
Uh oh.
The basic format that’s been proposed since around 2014 is that the science ‘side’ put their evidence in for animal models being ‘predictive’ as a referenced position paper and this is presented by a single spokesperson or champion. I spy at least four red flags there but let’s crack on.
The intention is that the other side will then submit their evidence for the impossibility of prediction and the whole thing will be judged by a panel of people with expertise in several areas including toxicology, philosophy and statistics. However, you cannot be a judge if you’re a scientist with a notable discovery that came from using an animal in an experiment, or have a relative or spouse who has ever received funding for conducting an experiment, for instance. They’ve had enough of experts.
This exempts a vast number of people with relevant technical knowledge of the topic being discussed from being judges. Even as advisors their directly relevant knowledge would be diluted by the presence of philosophers, physicists and even practitioners of ‘personalised medicine’ who are listed as required participants.
There are numerous other bits of small print but this is the gist and, the longer you read or think about it, the more problems emerge.
What’s the precise definition of ‘prediction’ they are using and why is it so tight? Why is that the thing being debated? Prediction of any kind isn’t really a thing in most animal modelling. Animal models are not used as a form of precognition, for instance, but might give a certain probability of safety and there are numerous possible meanings of the word depending on context. They might provide an analogy or signpost a successful approach, but it’s not intended as a precise blueprint.
For instance, I could predict that an activity is 96% safe based on the number of previous negative incidents, but I couldn’t say whether an individual would be in the 96% safe or 4% unsafe. I can say that walking into a road blindfolded has a 90% risk of injury, but you might be fine! The fact you can’t tell what will happen to an individual doesn't make the 90% part ‘junk science’. Prediction in this context is a probability, not a fortune-teller. Is that what the activists think we’re doing when we use an animal?
Another issue with the proposal is that it deals only with previously published materials, when the best data can be that generated especially for an inquiry. An example would be the IQ Consortium of pharma companies mentioned earlier. The pharma industry doesn’t particularly want to use animals and quite famously pumps vast amounts of money into finding alternatives to them. Yet, when it looked at whether animal safety translated to human safety and shared data on drug reactions collaboratively with one another, it discovered the following concordance:
|
Organ category |
Dog NPV (no negative reaction) |
|
Pulmonary |
96% |
|
Biochemical |
95% |
|
Renal |
95% |
|
Ophthalmology |
94% |
|
Haematology |
93% |
|
Cutaneous |
91% |
|
Musculoskeletal |
91% |
|
Cardiovascular |
91% |
|
Nervous system |
90% |
|
Liver |
88% |
|
Gastrointestinal |
76% |
A similar, effective, way of investigating animal use in research was undertaken by the UK’s centre for animal alternatives, the NC3Rs. It asked, since the globe’s regulators use two species of animal (one rodent and one non-rodent), to test drug safety, how much value does the second, non-rodent species add? Could it be replaced? Or dropped altogether?
The results were nuanced. Further areas of investigations were identified including potentialities for reducing the use of animals. But what the investigation most vividly demonstrated were the immense challenges in replacing animals without causing greater harm.
Even this one, small, subsection of animal use, parts of which could arguably be about prediction, doesn’t lend itself to a debate or a hearing that tries to present a side for or against.
So, what does debate settle?
It’s hard to think of any even basic question or premise that’s best settled by debate. Is the sky blue? Is grass green? It depends, doesn’t it? On complex atmospheric, soil and climatic conditions to name a few aspects.
What’s the point of hearing two cherry-picked presentations of when grass is green, or grass is not green? Especially when the question has so much nuance to the extent that the question itself is basically meaningless. We are far better off knowing why grass appears green or brown than asserting it’s one or the other.
This example also shows the importance of our understanding of the terms we are using. When we say grass do we mean Fescue lawn grass in rainy England or Californian scrub grass after a drought
Concerns
So, there are basic conceptual problems with debate as a means of determining facts. It’s not intended for that purpose so much as to persuade based on a narrative. Calling for a hearing doesn’t fix any of those problems and any process that forgoes data for debate isn’t scientific. This gives us our first issue:
1. Debate isn’t about truth, whereas science is
However, there’s much more. The proposed motion of the debate is:
The subject of the debate will be the position that animal models have insufficient predictive value for human response to perturbations that occur at higher levels of organization (e.g., human response to drugs and diseases) and the implication that the vast majority of animal use in science, in general, and research and testing, in particular, should cease.
This is strewn with important terms that need definition, and implicit assumptions that determine what is being discussed. Much as when asking whether grass is green or not, why must it be one or the other and not different colours in different contexts?
What does ‘predictive’ mean? What level of prediction is sufficient? What if things that fail the definition of prediction were yet sufficient?
The Public Assessment Report by the MHRA lists the animal data that supported the decision to allow human trials of the Oxford/AZ vaccine. It writes: “COVID-19 Vaccine AstraZeneca has been shown to be immunogenic in mice, ferrets, non-human primates (NHP) and pigs.” Is that prediction? Suggestion? One of many data points? Should we have disregarded the data because it fails some semantic test?
In examining the ability of animals to predict human outcomes, it presupposes that this is the purpose of the experiment. It’s a classic loaded question. The traditional example of this is asking for a yes or no answer to the question "Have you stopped beating your wife? Yes, or no?!"
Whether the respondent answers yes or no, they will admit to having a wife and having beaten her at some time in the past. These facts are presupposed by the question, and in this case are an entrapment, because it narrows the respondent to a single answer, and the logical fallacy of many questions has been committed.
We could further get into prediction being the wrong question if we’re genuinely interested in better medicines. The class of drugs with the highest failure rate in treating human disease is cancer drugs, but this is because they are much more likely to be tried in humans despite disappointing preclinical results. We could equally ask far more important questions about reproducibility and/or study design and reporting.
This gives us our second concern:
2. The key terms and definitions of the debate are both suspect and misrepresent the point of the research
All that said, let’s say that we do have a situation where we want to ask if animals predict human effects:
3. We have far better ways of settling this question
I’ve already mentioned the IQ Consortium and the NC3Rs, which are a pharma-led collaboration vehicle and a state-funded 3Rs initiative respectively.
Both allow data-sharing among expert stakeholders who might not ordinarily see it – academics, commercial competitors, regulators – but no one involved has any stake in the continued use of animals. Personally, and professionally, they just want to make sure that drugs are safe and effective. This is who you want supplying the data and doing the analysis.
Numerous papers and consortia have found similar values, including efforts by anti-vivisection campaigners that backfired, but ultimately these questions are best answered through the collaboration of experts who can bring to the table helpful insights, good data and a good faith attitude to discerning the facts. How far this is from a trial-style debate with peculiar rules and based on whatever data can be cherrypicked from the scientific literature.
Is there a reason 4? Yes and no.
There is one potential objection to these debates that I would not normally entertain because it’s a logical fallacy to reject a message simply because of the messenger. There are certainly people who will dismiss out of hand anything that comes from a particular source. We all do it to an extent but, if we’re honest, we are normally poorer for it.
Somebody in a red Donald Trump hat might dismiss the concerns of a trans activist as identity politics, or Guardian readers might dismiss Brexiteers as fools for rejecting the opinion of experts, but the truth is rarely as simple and polarised as all that. There are honest concerns in the points people on all sides raise that deserve honest answers.
Yet, the campaigners pushing for the debate have a decades-long track record of presenting doctored evidence with an ‘Instagram reality’ vibe. This puts us in a tricky spot. We should obviously be talking about this topic, and probably with these people, but not in this way.
So, where does this leave us?
In this case, the organisers of the debate could not have been any more persistently slippery over years and decades if they tried, and now they’re suggesting a ‘debate’ with rigged terms. The pseudoscience that lies behind both their worldview and the weird terms of their debate, designed to ensnare debaters in semantics, is deliberate. So too are the bad data sets, misused references and hateful character assassinations against scientists that activists seem to think are OK. The pseudoscientist Ray Greek thinks it’s ok to accuse cancer researchers of being driven by money and ego. This sort of irresponsible rhetoric has inspired celebrities such as Ricky Gervais and actor Peter Egan to jump in and publicly jeer at committed, hardworking scientists who are expected to simply turn the other cheek.
Politicians have been keen to jump in too – Lisa Cameron of the Scottish National Party leads the charge with EDM 175 - the latest of several Early Day Motions to call for science to be undermined, but smaller names in all the main political parties have accepted the false claims of activists without consulting a single credible source to see if they’re true. Moreover, they’re calling for a rigged debate on the basis of a conspiracy theory. It is unutterably irresponsible, the worst kind of populist politics and creates a safe space for misinformation-led vitriol to thrive.
Have you seen the data behind the science lecture that the campaigners claim demonstrates the grounds for concern? It’s just six compounds from an out-of-print book from 1990. Here they are:
It’s from this book.
I wish I was kidding. That’s not a dataset. This magnum opus of the anti-vivisection movement is practically written in crayon. The judge has a toy squeaky hammer as a gavel.
We need to talk about animal research, but this topic is nuanced, complicated and understanding it isn’t best achieved by polarised debates. We need to talk about what is being done to move away from animal models, and we need to continue talking about the utility of animal models, something that is constantly under review in scientific circles. But while we’re doing that, we also need to talk about why politicians and celebrities think it’s OK to punch down at medical researchers and use their platform to bully scientists whose work they don’t understand, without feeling the need to ask them about their research first.
Last edited: 21 April 2022 15:16
