Breastfeeding advocates have criticised the phrase ‘breast is best’ as mistakenly representing breastfeeding as a departure from the norm rather than the default for infant feeding. Breastfeeding mothers have an interest in representing breastfeeding as the default, for example, to counteract criticism of breastfeeding outside the home. This connects to an increasing trend to frame feeding babies formula as harmful, which can be seen in research papers, public policy and information presented to parents and prospective parents. (1) Whether we frame infant-feeding decisions in terms of harming or benefit, protection or risk matters because these distinctions are generally morally significant and thus (2) holding that those who decide to use formula ‘harm’, ‘risk harm’ to their babies or describing formula feeding as ‘dangerous’ is likely to contribute to guilt associated with formula feeding and thus to undermine the well-being of vulnerable women. It may undermine attempts to improve breastfeeding rates by leading women to reject information about health outcomes surrounding infant-feeding decisions. However, (3) these distinctions do not apply easily to infant-feeding decisions, in part because of difficulties in determining whether we should treat breastfeeding as the normative baseline for infant feeding. I show that neither the descriptive ‘facts of the matter’ nor moral or pragmatic considerations provide an easy answer before discussing how to respond to these considerations.
Statistics from Altmetric.com
This paper argues against a worrying trend that can have significant detrimental effects on the well-being of vulnerable women and points out a philosophically interesting aspect of early motherhood. The phrase ‘breast is best’ is controversial for many reasons, but some of the strongest criticism comes from breastfeeding advocates. This way of talking is claimed to mistakenly represent breastfeeding as a departure from the norm rather than the default for infant feeding. Breastfeeding mothers have an interest in representing breastfeeding as the default, for example, to counteract criticism of breastfeeding outside the home. I connect this issue to an increasing trend to frame infant formula as harmful in research papers, public policy and information presented to parents and prospective parents. I argue (1) whether we frame infant-feeding decisions in terms of harming or benefit, protection or risk matters because these distinctions are generally morally significant and thus (2) holding that those who decide to use formula ‘harm’ or ‘risk harm’ to their babies or describing formula feeding as ‘dangerous’ is likely to contribute to guilt associated with formula feeding and thus to undermine the well-being of vulnerable women and be counterproductive with respect to increasing breastfeeding rates. However, (3) these distinctions do not apply easily to infant-feeding decisions, in part because of difficulties in determining whether we should treat breastfeeding as the normative baseline for infant feeding. I show that neither the descriptive ‘facts of the matter’ nor moral considerations provide an easy answer before discussing how to respond to these considerations.
Harms of formula versus benefits of breast feeding
Many breastfeeding advocates criticise the phrase ‘breast is best’. ‘The Analytical Armadillo’, a popular International Board Certified Lactation Consultant blogger, argues: “Breastmilk is the bog standardi, normal substance a human infant is built to consume—there’s nothing best or superior about it”.1 Calvert echoes this in the practitioner journal, Nursing Children and Young People: “Breastfeeding is the biological norm, the way that humans have evolved to feed their children. Breast is not ‘best’, it is normal”.2
Such discussions often refer to Wiessinger’s influential paper, ‘Watch Your Language’. Wiessinger argues that breastfeeding should be seen as standard and that formula feeding should therefore be seen as risky or harmful: “The truth is, breastfeeding is nothing more than normal. Artificial feeding, which is neither the same nor superior, is therefore deficient, incomplete, and inferior”. Later, she states: “(A mother) is less likely to use artificial baby milk ‘just to get him used to a bottle’; if she knows that the contents of that bottle cause harm”.3 On this model, holding that breastfeeding is normal leads us to see formula feeding as harmful. This in turn may lead us to classify a woman who decides to use formula as harming her child.
Wiessinger’s paper has had a significant impact on discussion of infant feeding. For example, a pair of adverts from the National Breastfeeding Campaign show pregnant women taking part in a log-rolling competition or riding a mechanical bull, followed by the slogan: “You wouldn’t take risks before your baby’s born. Why start after?”4 5; in a paper summarising the scientific evidence of differences in health outcomes in Reviews in Obstetrics and Gynaecology, Stuebe explicitly defends framing discussion in terms of ‘the risks of formula’, appealing to Wiessinger6 ,ii; public information pages on the La Leche League UK claim that ‘formula feeding increases the risks of’ a long list of illness7; the ‘toolkit for local authorities’ produced by Public Health England and Unicef Babyfriendly UK, claims that “Not breastfeeding increases a baby’s risk of obesity, diabetes, respiratory infections, gastroenteritis, ear infections, tooth decay, allergic disease and sudden infant death syndrome”.8
How we frame infant-feeding decisions matters because the most intuitively plausible morality draws deontological distinctions. When my behaviour affects another, we do not just care about how much their well-being has gone up or down. It matters whether they have suffered a harm or merely failed to receive a benefit. It matters whether this is something I have done or merely allowed. Some philosophers reject the moral relevance of these distinctions, but, as I have argued elsewhere, any moral theory that does not recognise such deontological distinctions has radically counterintuitive implications.9
Different ways of categorising the decision to use formula have different moral implications. These are particularly serious if we categorise feeding a baby formula as ‘harming’ or ‘doing harm’. Our ordinary moral reasoning recognises much stronger constraints against harming than failing to prevent harm or failing to bestow a benefit. Harming may require much stronger justification if it is to be permissible on grounds of avoidance costs to the agent or the greater good; harming leads to much stronger reactive attitudes such as blame and guilt if unjustified. If we frame formula feeding as harming—especially if at the same time we hold that the difference in outcomes is significant—we imply that women who formula feed without extremely strong justification are liable to blame and guilt.
Other formulations that fall short of explicitly describing the mother as harming her child also have important moral implications. Sometimes, they are taken to imply that the mother is risking harming her child: this applies to talk about the ‘risks’ or ‘dangers’ of formula feeding. This language describes what the mother does as dangerous and thus invokes the constraint against harming. In other cases, as when formula itself is described as ‘harmful’ or as ‘increasing risks of harm’, a weaker, but still significant, constraint is invoked. There are much stronger constraints against harming than merely refraining from preventing harm. However, there are stronger constraints against refraining from preventing harm or risk of harm than against refraining from benefiting. You might think that if I see a nearby stranger about to fall over and hurt himself, I am required to help him, even at some cost to myself, but I have no duty to put myself out to help him acquire extra pleasurable experiences. Requirements to prevent harm or the risk of harm are particularly strong in the parental case where there is a duty of protection. Parents are seen as having extremely strong duties to protect their children from risks of harm even at extremely substantial cost to themselves. So even ascriptions of harm to the product itself rather than the mother’s behaviour have moral implications.
Given the detrimental effects that shame surrounding formula use can have on the well-being of new mothers and their neonates, we have strong reasons to avoid the unjustified use of morally loaded terms to describe infant-feeding decisions. There is significant sociological evidence connecting decisions to use formula and feelings of shame, guilt and failure.10 These feelings in themselves can represent a significant cost to the well-being of a vulnerable new mother. Moreover, there is evidence that women who use formula may conceal their practices from midwives and other health professionals.11 If women cannot be honest with their health professionals, neither they nor their neonates will receive thorough healthcare. The use of morally loaded terms can be expected to contribute to maternal guilt and shame surrounding infant-feeding decisions, and it also gives the impression that such guilt and shame is appropriate. If guilt and shame is seen as appropriate, then its effects on maternal well-being may be wrongly dismissed as morally unimportant.
The use of morally loaded terms may also be counterproductive in terms of increasing breastfeeding rates. Lee and Furedi report that perceived pressure to breastfeed can lead women to ‘come to distrust professionals, and become sceptical about the value of professional knowledge and advice’.11 Telling mothers that they are harming their babies can be expected to have just this effect. A recent study found that pregnant women were more inclined to reject risk-based text on infant-feeding decisions than the same text framed in terms of benefits.12
Comparative accounts of harm versus benefit: is there a moral baseline for infant feeding?
Wiessinger appeals to an allegedly standard use of language surrounding health to argue that we should treat breastfeeding as the default and formula feeding as deficient and dangerous. She states: “Health comparisons use a biological, not a cultural, norm, whether the deviation is harmful or helpful. Smokers have higher rates of illness; increasing pre-natal folic acid may reduce fetal defects”.3 Wiessinger claims that our discussions of infant-feeding choice invert this standard pattern in which the biological norm is treated as the default or baseline for comparison. On this view, we talk about the harmful effects of smoking, not the benefits of giving up smoking, because being smoke free is the biological norm. In contrast, when we say that breastfed babies are ‘smarter’, we treat formula as the default.3
It might be thought that Wiessinger’s main point is that breastfeeding is the biological norm and thus should be taken as the departure point when describing infant-feeding decisions. Perhaps this could be achieved equally well without using morally loaded concepts: perhaps we could talk about formula as less good than breastfeeding in a morally neutral way. I have no objection to morally neutral reframing of breastfeeding as the norm. My interest in this paper is in the widespread use of harm and risk-based language, which has been influenced by Wiessinger. This use of harm and risk-based language is described above.
Even if breastfeeding is the biological norm, it is far from obvious that it should be the moral baseline from which the morally loaded calculations of harm and benefit are calculated. To illustrate why, consider the following thought experiment. In some species of spider, the biological norm is for baby spiders to eat their mother as their first meal. Suppose that such spiders evolved to become moral agents. Spider mothers who gave their spiderlings artificial food instead of the biological norm of mother’s meat should not be classified as harming their children. We cannot simply read off the moral baseline from the biological norms.
Of course, breastfeeding is generally very different from being consumed by baby spiders. However, they raise somewhat similar problems for using the biological norm as the moral baseline for harm/benefit classifications. Breastfeeding requires the mother’s body and agency. Breastfeeding is doing something. This claim is not undermined by the ‘breast crawl’ in which a baby, placed on the mother’s abdomen after birth, instinctively moves to their mother’s breast and begins breastfeeding.13 14 The breast crawl requires someone to put the baby on the mother’s stomach and for the mother to maintain the right position. In general, breastfeeding is a full exercise of agency in which the mother plans in advance to purposely position herself and her baby so that the baby is able to feed, monitors whether her baby is successfully feeding and makes adjustments if this is not happening. This is surely doing something. Moreover, it is doing something with her body that is intimate, time-consuming and sometimes difficult.
Because breastfeeding deeply implicates the mother’s body and agency, positioning breastfeeding as the moral baseline is problematic even if it is the biological norm. To do so takes the mother’s body and agency for granted. It does not fit with our use of the concepts of harm and benefit in other situations. In general, we think of harming someone as making them worse off than a neutral state of non-interference. Breastfeeding cannot be characterised as a neutral state of non-interference.
One response to these considerations is to adopt an agency-based account of harming. On an agency-based account, an agent counts as harming if and only they make another person worse off than that person would have been if the agent had done nothing. The agency-based account fits with the thought that harming someone involves interfering with their well-being. If they would have been just as badly off if you had done nothing at all, you cannot count as harming them.15 16
The agency-based account is normally proposed as an account of when an agent counts as harming, but agency-based accounts of what it is to suffer harm are possible. An agent-relative agency-based account takes as the baseline what would have happened if a particular agent had done nothing. You suffer harm (relative to me) if and only if you would have been better off if I had done nothing. An agent-neutral agency-based account takes as the baseline what would have happened if no agent had acted: you suffer harm if and only if you would have been better off if no one had done anything.
Neither agency-based account of harm applies well to infant-feeding decisions. They suffer the mirror image of the problem that the biological-norm account faced. While biological-norm accounts took for granted the mother’s exercise of her agency, the agency-based accounts treat anything the mother does as a bonus. Because, typically, the infant relies on its mother (or other caregivers) for survival, on the agency-based account, the baseline is death. Anything better than death is a benefit.
The problems with both the biological-norm account and the agency-based accounts are caused by the special relationship between the mother, as one of the primary caregivers, and the infant. Because of the infant’s dependence on its primary caregivers, we cannot make sense of a baseline level of well-being that is independent of their behaviour. In the maternal case in particular, physiological processes such as breastfeeding involve interaction between the infant and the mother. The biological norms of the infant involve the mother’s body and agency. Thus, any attempt to understand the notion of harming versus benefiting by comparison with a neutral baseline state will run into problems. There is no such thing as a neutral state of non-interference when it comes to mother and infant. Kingma and Woollard argue that because of the absence of a neutral state of non-interference between mother and infant in pregnancy, key ethical concepts like harming and benefiting do not apply normally to pregnant women’s interactions with their foetuses (E Kingma, F Woollard, Can you do harm to your fetus? Pregnancy, physical indistinctness, and difficult deontological distinctions). After birth, the mother and child are physically, but not socially or physiologically, distinct. There is still no neutral state of non-interference. So it is not surprising if we find this phenomenon that has been identified in pregnancy persisting into the postpartum period and applying to infant-feeding decisions.
Non-comparative accounts of harm
Can we understand the notions of harming and suffering harm without a baseline state for comparison? The most promising approach seems to be to combine a ‘process account of harming’ with an intrinsic harm account. Intrinsic accounts of harm hold that a person suffers harm if they are in a state that is bad in its own right. The state does not need to be bad in comparison with another possible state. To suffer pain is to suffer harm even if there is no baseline state where you would not suffer that pain. The term ‘process accounts of harming’ is intended to include all accounts that say whether you do harm depends on whether the right kind of process connects your behaviour and the harm. Foot, Draper, Woollard and Barry and Øverland offer such accounts.9 17–21 Consider a clear case of doing harm, pushing someone off a cliff. When I push someone off a cliff, I throw out my arms, connecting with his body, his body falls off the cliff, his leg strikes the rock and the leg breaks. The right kind of process connects my behaviour and the harm to him. I count as harming him. If he is already falling off the cliff and I just refrain from doing anything, there does not seem to be a similar process connecting my behaviour and the harm. I count as merely allowing him to be harmed. Combining the intrinsic account of harm and the process account of harming gives us the view that an agent harms another if and only if the right kind of process connects her behaviour to an intrinsically bad state. This might seem like the most promising approach to infant-feeding decisions because it does not require a comparison with a neutral baseline.
Unfortunately, we cannot simply appeal to process plus intrinsic harm accounts to solve the problems in applying the harm/benefit distinction to infant feeding. Even if this approach can be made to work, it does not give a straightforward answer to the question of whether to speak of the harms of formula or the benefits of breastfeeding. It requires us to look at each potential consequence in detail and identify the mechanisms at work. Formula will count as harmful only when there is an appropriate mechanism connecting it to an intrinsic harm. If there were a difference in rates of some disease due to, for example, microbes present in breastmilk that support the immune system, this would not count as a harm from formula according to the process/intrinsic harm approach. In such a case, there is no positive process connecting the use of formula and the harm. The use of infant formula is connected to the disease solely through the absence of a mechanism that would have protected the baby if the mother had breastfed. It is analogous to the case where I did not save the person from falling off a cliff: my behaviour is connected to the harm only through the absence of a protective act.
Moreover, intrinsic harm accounts have problems of their own outside of discussion of infant-feeding decisions. If the intrinsic harm account is not a good account of harm in general, then it is not a good account of harm associated with infant-feeding decisions. Arguably, the intrinsic harm is not a good account of harm in general. Death is a paradigm harm. Any account that cannot recognise death as a harm is extremely implausible. However, the intrinsic harm account struggles to correctly classify death as a harm. Death itself may not be intrinsically bad.22 Early death seems to be bad principally because it causes us to miss out on the good things we could have if we were alive. So it looks as if we need some kind of comparative account to correctly classify death as a harm. There may be other harms, such as losing a physical ability, which are primarily bad because of what they cause us to lose and thus are not captured by the intrinsic harm account. Moreover, we might think that a person can suffer even quite a significant amount of pain without it being appropriate to speak of them as suffering harm. Coughs, colds, stomachaches and so on seem like part and parcel of what it is to be human. These considerations suggest that we should reject non-comparative accounts of harm in general and adopt a comparative account instead.
However, suppose that we could resolve these general worries about the intrinsic account of harm and that we pick out an active mechanism leading from formula feeding to an intrinsic harm. There are still difficulties in applying process accounts of harming to the infant-feeding case due to the unique relationship between the mother and infant. Process accounts pick out the connection between the agent’s behaviour and a harm that is needed for the agent to count as interfering with the other. Strong constraints against interfering in this way are normally possible because the agent has the option of not interfering. In general, the alternative to having such processes link my behaviour to a bad effect on another is for there to be no processes linking my behaviour to an effect on another. In the mother–infant case, non-interference with the infant is not an option. Because strong constraints against harming presuppose the option of not interfering, there are much weaker constraints against non-standard harmings where there is no option of non-interference. For example, costs to the agent, which would not justify a standard harming, may justify a non-standard harming. So even if using formula counts as harming on a process account, it will be importantly different from standard harmings. Talking about ‘harming’ in this context may be misleading (E Kingma, F Woollard, Can you do harm to your fetus? Pregnancy, physical indistinctness, and difficult deontological distinctions).
This section has focused on non-comparative accounts of harming. Analogous non-comparative accounts of benefiting hold that an agent benefits another if and only if the right kind of process connects her behaviour to an intrinsically good state. Given that it seems desirable to have symmetrical accounts of benefiting and harming, the considerations raised in this section against the non-comparative account of harming also speak against a non-comparative account of benefiting.
Summary of discussion of accounts of harm/benefit
The idea of the normative baseline, and the related notions of harming versus allowing harm or failing to benefit, and risks versus benefits, do not apply easily to infant-feeding decisions. Fixing on the biological norm ignores the role of the mother’s agency in breastfeeding. Treating the baseline as ‘doing nothing’ treats anything the mother does as a bonus. Non-comparative accounts are implausible as general accounts of harm and benefit and, in any case, do not provide straightforward answers when it comes to infant feeding. The descriptive facts of the matter do not provide an easy answer.
Moral and pragmatic considerations
Those who breastfeed and support breastfeeding have reason to want to emphasise breastfeeding as the default method for infant feeding. Seeing breastfeeding as the default may be helpful in counteracting negative attitudes towards breastfeeding. Many mothers still feel embarrassed breastfeeding outside the home. Seeing breastfeeding as the default may be useful in arguing that breastfeeding should be supported even in non-ideal situations. Calvert has written movingly on support for breastfeeding on the children’s wards in hospitals. She connects the ‘Breast is best’ rhetoric to an attitude that breastfeeding is an optional extra which mothers cannot expect busy health professionals to make significant efforts to facilitate.2 Seeing breastfeeding as the default may help to encourage mothers who are failed by insufficient support to become angry and to demand better support. If we see ourselves as having lost out on the ‘best’, then we might not see this as a big problem. If we think that institutional priorities lead to us being unable to do the ‘bog standard’,1 then we are more likely to be galvanised into action. These considerations speak against framing infant-feeding decisions in terms of the benefits of breastfeeding.
On the other hand, those using formula, and those who care about them, have reason to resist characterisation of breastfeeding as the default where this in turn unjustly portrays those who use formula as harming their babies or leaving them vulnerable to harm, adding to the already substantial guilt and shame that can accompany formula feeding.10 11 The connections between characterising formula as harming, guilt and shame and the detrimental effects on the well-being of mothers and their neonates are discussed in detail in section 1.
In addition, we should not be misled by these characterisations into thinking that this is a simple case of the interests of one group of mothers (formula feeders) versus the interests of a completely separate group (breastfeeders): one and the same mother may both use formula and breastfeed, either at the same time or at different times. Thus, one and the same mother may require support for breastfeeding and protection against guilt and shame for using formula.
So there are concerns about both framing infant-feeding decisions in terms of the benefits of breastfeeding and in terms of the risks or harms of formula. We might try to avoid these issues by using neutral language such as ‘difference’, accompanied by clear information about the outcomes presented non-comparatively: that is, “There are differences in rates of disease X between formula fed and breastfed babies. A typical breastfed baby has a 1/N1 chance of suffering disease X; a typical formula fed baby has a 1/N2 chance of suffering disease X”. However, such formulations may be long-winded and off-puttingly technical, presenting a barrier to providing key public health messages.
Neither the descriptive facts of the matter nor moral or pragmatic considerations provide an easy answer about how we should characterise maternal decisions about infant feeding.
Two caveats: third parties and extreme cases
I have argued that the idea of the baseline state of non-interference, and the related notions of harming versus allowing harm, and risks versus benefits, do not apply easily to maternal decisions about infant feeding. However, the issues with the baseline that I have picked out apply mainly to the mother’s or primary caregivers’ interactions with the infant. Thus, the arguments in this paper do not show that we cannot find a neutral state of non-interference to evaluate third parties’ interactions with the mother–infant dyad.
I hold that the unique nature of the mother–child relationship means that many key ethical concepts may not apply to interactions between mothers and their children. Thus, the further conclusion that harming language in general is inapplicable to this context would be an interesting result, rather than a reductio for me. Nonetheless, one can accept the arguments of this paper without being committed to this stronger conclusion. I have shown that we need to be extremely careful in using harming language in the maternal context. We do not yet have a clear understanding of how to apply the harm/benefit distinction and related moral concepts to mother–child interactions. Given this, and given the moral force of harm language and the pragmatic considerations, we should not use harm language to describe a mother’s decisions to use infant formula. However, someone may still think that an appropriate account of how to apply the harm/benefit distinction to maternal behaviour is out there. There may be cases such that (1) it is clear that any appropriate account of harm will count them as harms; (2) the relevant alternatives only minimally involve the mother’s body and agency; (3) there are strong pragmatic reasons for using harm language that heavily outweigh any reasons not to use harm language. The arguments in this paper do not show that we should not use harm language in such cases.
When it comes to descriptions of maternal behaviour, we should reject the assumption that there has to be a single appropriate default for infant feeding. Breastfeeding is normal and should not be stigmatised or seen as a lifestyle choice that can only be accommodated under ideal circumstances. The phrase ‘breast is best’ should be avoided. But we should not treat breastfeeding as a baseline in a sense that implies that women who formula feed are harming their babies. Extreme care should be taken before using morally powerful terms such as ‘risk’, ‘harm’ and ‘danger’. Where possible, neutral terms such as ‘difference’ should be used, accompanied by clear information about the outcomes presented non-comparatively.
I would like to thank three anonymous referees and Emma Veitch for extremely helpful comments.
↵i ‘Bog standard’ is British slang meaning ‘utterly basic or ordinary’.
Contributors This paper is the sole work of the author.
Funding This paper is associated with a research project funded by Program for Research, Development and Innovation Oriented to Societal Challenges, Ministry of Economy in Spain (grant no. FFI2016-77755-R).
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No unpublished data are used in the study.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.