Hooks et al. identified many important problems in the methodology and interpretations of microbiota-gut-brain (MGB) research. Here I focus on one facet of MGB interpretations: the tendency to explain experimental effects by appealing to the microbiome as a whole. The authors allude to the problem of citing whole microbiomes – entire ecosystems or communities of microbes – as causes of behaviour and brain states. Yet a positive account of more appropriate explanations for these findings is not offered. I propose that this explanatory inadequacy is attributable to a failure of proportionality of explanation, given the current state of research in the field.
Proportionality is an explanatory tool that philosophers of science use to identify the most appropriate causal explanation when assessing scientific findings. Proportional explanations must not be too broad that they include aspects that are irrelevant to the effect produced, and not so narrow that important causal information is omitted (Woodward Reference Woodward2010). For example, if a pigeon is trained to peck whenever a red target is presented, then a disproportionate explanation for pecking behaviour would be the presentation of a coloured target. Citing a coloured target as a cause is too broad as it encompasses factors irrelevant to the effect, namely, other (non-red) coloured targets. Explaining pecking behaviour as caused by the presentation of a scarlet target is too narrow, as it excludes other variations of red that would also elicit this behaviour. A proportional explanation for pecking behaviour is simply the presentation of a red target (Yablo Reference Yablo1992).
In MGB research, microbiomes are cited as causes in one of two ways. The first is explicit reference to the whole microbial community as causing a brain or behavioural state (for examples, see sect. 7.1, para. 2). The second is when the microbiome as a community is implicitly suggested by reference to “dysbiosis” – a concept that indicates that the microbiome is “imbalanced,” often implying decreased microbial diversity (for examples, see sect. 7.1, para. 3). To convincingly explain brain or behavioural states with microbiomes, the whole microbiome must be shown to be causally relevant. Researchers must demonstrate that manipulating microbiomes as a whole produces effects that cannot be reduced to just some of their component parts – via manipulations of one or a few microbes alone.
Most of the evidence presented regarding MGB research, however, does not support the thesis that the entire microbiome is causally involved. Manipulations using probiotics (Supplementary Table 2 in Hooks et al.) introduce one or at most a few species of bacteria – a tiny component part of the microbiome in its entirety. Making causal claims about microbiomes in these instances is akin to claiming that a coloured target caused the pigeon to peck. The implication is that entire communities of microbes exert a causal influence on behaviour, yet experimental evidence suggests single or a cluster of few microbes as proportional causal explanations.
Some researchers object to reductionist explanations of this kind by claiming that interventions using one or a few strains of bacteria make significant changes to the microbial community (e.g., Arnold et al. Reference Arnold, Roach and Azcarate-Peril2016). However, the burden of proof is on researchers to demonstrate (1) these large-scale changes occur after single species introductions, and (2) such changes are responsible for the brain or behavioural state observed, as opposed to the microbial introductions acting directly. Supporting (1) and (2) seems unlikely, as a review of probiotic treatments on human microbiota revealed no overall effects on α-diversity, richness, or evenness (Kristensen et al. Reference Kristensen, Bryrup, Allin, Nielsen, Hansen and Pedersen2016).
Probiotics are not the only interventions made in MGB research. Other methods include the use of antibiotics to “deplete” the microbiome and transplanting faecal matter from humans or other rodents into germ-free rodents. It could be argued that these more strongly support whole-microbiome explanations, as faecal matter is thought to contain entire microbial communities, which are completely transferred. However, faecal microbiomes are not synonymous with gut microbiomes, as many microbes may not make it through the digestive tract (sect. 5, para. 3). This aside, experiments transplanting whole microbiomes via faecal matter transplants or eliminating gut microbiomes using antibiotics still run into trouble when it comes to proportional causal explanations.
To return to Yablo's pigeon, imagine you are an experimenter trying to understand what makes the pigeon peck. You present a sequence of coloured targets, and the pigeon pecks. What do you conclude? One explanation is that coloured targets make the pigeon peck. Although this accurately describes the results of the experiment, an obvious follow-up would be to try and eliminate some of the colours by presenting only component parts of the original sequence. Coloured targets may be a useful “black box” or “placeholder” explanation in this case, but any critical reader would see that follow-up experimentation is needed to identify a better causal explanation.
So are microbiome explanations simply, as Hooks et al. put it, “placeholder claim(s)” for future causal claims, typical of explanations made in a younger field? I do not believe so. In the early days of gene research, heritability estimates served as a “black box” for genetic explanations for many traits. Yet reference to “the genome” as a causal explanation did not permeate this field. Instead, causal claims about traits being primarily “genetic,” later displaced with “gene for” terminology, dominated (Griffiths & Stotz Reference Griffiths and Stotz2013). This is akin to explaining brain or behavioural states as primarily “microbial” or referring to a “microbe for” certain traits.
In MGB research, a black-box strategy may be implicit for some – yet is not obvious to the majority, as community-level explanations remain rife within the literature and in the popular press. Extending Hooks et al.’s suggestions for social scientific study (sect. 8, para. 3), investigation as to why community-level explanations have dominated the field is warranted. In the meantime, researchers should strive for greater precision when communicating their research, and considering the proportionality of causal explanations will be useful for this purpose.
You have
Access
Hooks et al. identified many important problems in the methodology and interpretations of microbiota-gut-brain (MGB) research. Here I focus on one facet of MGB interpretations: the tendency to explain experimental effects by appealing to the microbiome as a whole. The authors allude to the problem of citing whole microbiomes – entire ecosystems or communities of microbes – as causes of behaviour and brain states. Yet a positive account of more appropriate explanations for these findings is not offered. I propose that this explanatory inadequacy is attributable to a failure of proportionality of explanation, given the current state of research in the field.
Proportionality is an explanatory tool that philosophers of science use to identify the most appropriate causal explanation when assessing scientific findings. Proportional explanations must not be too broad that they include aspects that are irrelevant to the effect produced, and not so narrow that important causal information is omitted (Woodward Reference Woodward2010). For example, if a pigeon is trained to peck whenever a red target is presented, then a disproportionate explanation for pecking behaviour would be the presentation of a coloured target. Citing a coloured target as a cause is too broad as it encompasses factors irrelevant to the effect, namely, other (non-red) coloured targets. Explaining pecking behaviour as caused by the presentation of a scarlet target is too narrow, as it excludes other variations of red that would also elicit this behaviour. A proportional explanation for pecking behaviour is simply the presentation of a red target (Yablo Reference Yablo1992).
In MGB research, microbiomes are cited as causes in one of two ways. The first is explicit reference to the whole microbial community as causing a brain or behavioural state (for examples, see sect. 7.1, para. 2). The second is when the microbiome as a community is implicitly suggested by reference to “dysbiosis” – a concept that indicates that the microbiome is “imbalanced,” often implying decreased microbial diversity (for examples, see sect. 7.1, para. 3). To convincingly explain brain or behavioural states with microbiomes, the whole microbiome must be shown to be causally relevant. Researchers must demonstrate that manipulating microbiomes as a whole produces effects that cannot be reduced to just some of their component parts – via manipulations of one or a few microbes alone.
Most of the evidence presented regarding MGB research, however, does not support the thesis that the entire microbiome is causally involved. Manipulations using probiotics (Supplementary Table 2 in Hooks et al.) introduce one or at most a few species of bacteria – a tiny component part of the microbiome in its entirety. Making causal claims about microbiomes in these instances is akin to claiming that a coloured target caused the pigeon to peck. The implication is that entire communities of microbes exert a causal influence on behaviour, yet experimental evidence suggests single or a cluster of few microbes as proportional causal explanations.
Some researchers object to reductionist explanations of this kind by claiming that interventions using one or a few strains of bacteria make significant changes to the microbial community (e.g., Arnold et al. Reference Arnold, Roach and Azcarate-Peril2016). However, the burden of proof is on researchers to demonstrate (1) these large-scale changes occur after single species introductions, and (2) such changes are responsible for the brain or behavioural state observed, as opposed to the microbial introductions acting directly. Supporting (1) and (2) seems unlikely, as a review of probiotic treatments on human microbiota revealed no overall effects on α-diversity, richness, or evenness (Kristensen et al. Reference Kristensen, Bryrup, Allin, Nielsen, Hansen and Pedersen2016).
Probiotics are not the only interventions made in MGB research. Other methods include the use of antibiotics to “deplete” the microbiome and transplanting faecal matter from humans or other rodents into germ-free rodents. It could be argued that these more strongly support whole-microbiome explanations, as faecal matter is thought to contain entire microbial communities, which are completely transferred. However, faecal microbiomes are not synonymous with gut microbiomes, as many microbes may not make it through the digestive tract (sect. 5, para. 3). This aside, experiments transplanting whole microbiomes via faecal matter transplants or eliminating gut microbiomes using antibiotics still run into trouble when it comes to proportional causal explanations.
To return to Yablo's pigeon, imagine you are an experimenter trying to understand what makes the pigeon peck. You present a sequence of coloured targets, and the pigeon pecks. What do you conclude? One explanation is that coloured targets make the pigeon peck. Although this accurately describes the results of the experiment, an obvious follow-up would be to try and eliminate some of the colours by presenting only component parts of the original sequence. Coloured targets may be a useful “black box” or “placeholder” explanation in this case, but any critical reader would see that follow-up experimentation is needed to identify a better causal explanation.
So are microbiome explanations simply, as Hooks et al. put it, “placeholder claim(s)” for future causal claims, typical of explanations made in a younger field? I do not believe so. In the early days of gene research, heritability estimates served as a “black box” for genetic explanations for many traits. Yet reference to “the genome” as a causal explanation did not permeate this field. Instead, causal claims about traits being primarily “genetic,” later displaced with “gene for” terminology, dominated (Griffiths & Stotz Reference Griffiths and Stotz2013). This is akin to explaining brain or behavioural states as primarily “microbial” or referring to a “microbe for” certain traits.
In MGB research, a black-box strategy may be implicit for some – yet is not obvious to the majority, as community-level explanations remain rife within the literature and in the popular press. Extending Hooks et al.’s suggestions for social scientific study (sect. 8, para. 3), investigation as to why community-level explanations have dominated the field is warranted. In the meantime, researchers should strive for greater precision when communicating their research, and considering the proportionality of causal explanations will be useful for this purpose.