On the cover of Time magazine, there's a three-year-old standing on a stool, breastfeeding. A combination of his height, and maybe the stool, means that he looks about seven. Three is not that old, in the world of "extended" breastfeeding, but seven is giving it some by anybody's standards. The mother, Jamie Lynne Grumet, said, "I understand some of the breastfeeding advocates are actually upset about this," which is most probably true. There's a tipping point, in the world of breastfeeding advocates: they are evangelical to the point of dogmatism about the desirability of breastfeeding an under-two, go a bit quiet between two and three, then back right off, on the tacit understanding that extended breastfeeders make them all look weird.

I'm guessing here, but I think the anxiety is that once the child looks like a proper little person, with proper teeth and enzymes, you enter a world of confusion about who needs what from whom; the kid doesn't need the milk, clearly. At this point it is pure attachment, pure love. Who needs it expressed this way, the mother or the child? I personally don't care. But if I were really invested in breast milk as liquid gold, I can see that I wouldn't want that to be the question people were asking.

I could not have loved breastfeeding more if I'd been brainwashed; I experienced it as a kind of hallucinogenic experience. A bit like taking an E. But I also had the strong suspicion that the claims made for its benefits – the higher IQ, the protection against obesity, the superior bonding, the warding off of disease both now and for ever, both for baby and for mother – were mostly bogus. A lot of the reasoning seemed syllogistic (babies born into low-income families end up fatter; low-income mothers breastfeed less than high-income mothers; therefore breastfeeding prevents obesity) or frankly lame. I knew a lot of mothers who formula fed; they didn't seem to love their babies less. When I wrote a book about pregnancy, I included some of this lameness, while underlining the fact that, speaking for myself, I didn't care whether or not the health benefits were real, I'd do it again even if it made the baby's IQ go down.

Two months later, I was on my way to chair a discussion for the Royal College of Midwives when I got a call from one of their press officers: she said there might be a demonstration against me outside the venue. "What do you mean?"

"Oh, nothing huge – there were just some people who argued against you being there, and they might… demonstrate."

"Are they going to egg me?"

"You're obviously a really sensitive person," she said.

"I'm not sensitive," I countered, although if I'm honest I did not relish the idea. "It will just affect what I'm going to wear."

Nobody demonstrated against me that night, but someone did approach me in the meeting saying they'd appreciate a retraction for the paperback. "Retract what?" I asked. Just retract. Stop saying what you're saying, because breastfeeding really is better.

A year later, a conference on infant feeding included the American academic Joan B Wolf, who conducts a rigorous, close-range examination of the science behind pro-breastfeeding advice in Is Breast Best? She concludes that the case for breast milk is hyperbolic. "The majority of studies have demonstrated that there's a relationship between breastfeeding and better health. But whether this relationship is causal has never been established." The only discernible benefit was in reducing instances of gastroenteritis, thanks to a specific agent found only in human milk, secretory immunoglobulin A (or SIgA). Again, there was the threat of a demonstration against her, this time outside the British Library, which would have been a bit embarrassing (you can't demonstrate against someone collating study results. What are you going to put on your placard? Down with Meta-Studies?). But again, it didn't happen.

This struck me as a bizarre place to have arrived at; where even to talk about the evidence behind the benefits of one type of infant feeding over another is heretical. The questioning of the orthodoxy is taken as a direct attack on babies. And who would attack a baby? Only a vile lunatic; so the temperature of the debate is often quite high from the start. Charlotte Faircloth, a sociologist who did her PhD on "full-term" breastfeeding in London and Paris (I think of her as the George Orwell of lactation), comments mildly: "I think, weirdly, not being a mum was really a useful thing for me while I was doing my research, because I wasn't a them or an us. It allowed me to ask some basic questions that I wouldn't have been able to otherwise. Everything has got very heated, and very moralised. How you feed your kids is no longer a personal decision. There's this idea that you can breastfeed your way out of poverty, or if you don't breastfeed your kid's going to be fat or have a low IQ…" She pauses. "It's all got a bit out of hand."

In some ways, with the distance afforded by now having two fully weaned children, I understand it: the act of breastfeeding – that is, sitting in Starbucks half-naked – is taken to be such a transgression of social norms that in order to make women feel OK about it, there has to be a degree of overstatement as to how beneficial it all is.

But there are some women daring to challenge the idea that breast is best. Repudiating "attachment" parenting, French feminist Elisabeth Badinter argues that breastfeeding – along with natural labour, co-sleeping and giving up work – all amount to an ideal of motherhood that effectively subjugates a woman's professional, sexual, spousal, adult identity. "If 24 hours a day the woman is reduced to her role as a nursing animal, even putting the child in the bed between the father and the mother, the father is completely put aside," she said recently, "This is very hard for men, and I think the child becomes a factor in the separation of the couple. Breastfeeding a few weeks, sometimes a few months, OK. But when it's recommended that you breastfeed your child for one year – six months exclusively, with nothing else, day and night, on demand – there are obviously consequences for a couple."

Badinter's beef with the idea of "total motherhood" has caused more of a furore in the US than it did in France, where her main opponents were Greens (she laid some of the blame for this essentialist maternal model at the door of "ecologists" with their non-disposable nappies). In the US, where arguably women have been in thrall to this idea of perfect motherhood (or "Borg-parenting", as the New York Times put it) for longer, the idea that motherhood and feminism might be at odds has been greeted in some predictable and some comically unexpected ways.

A number of commentators have said that, in fact, the right to breastfeed was a victory for feminism, in wresting the care of their babies from a professional, medicalised elite. Others have conceded that the struggle for perfection and unanimity in any direction – towards breastfeeding or away from it – is necessarily bad for women, removing their personal agency. (Which Bandinter doesn't argue with: "Let women do what they want!" she says. "There are women for whom breastfeeding is a true pleasure. It's very good for them and it's very good for the baby. But to breastfeed a baby if the mother herself doesn't like it? It's a catastrophe. The decision to breastfeed is an intimate and private decision. No one should be able to interfere.")

But there are certainly those, for instance LaShaun Williams, who argue that Badinter is right – motherhood is at odds with feminism, and this is a good thing, since when we finally accept the primacy of the maternal role, it shows up feminism for the sham it is. "We should question," she writes, "why so many of us are working – single and married women alike. Is it because we bought the feminist lie that we don't need a husband? Is it because we want to prove to the world that we are worth something? Or is it to live in a ritzy neighbourhood and drive an Audi Q7? When we bring children into this world we also agree to sacrifice parts of ourselves." In this version of events, just by wanting financial agency, women have succumbed to avarice, and a feminist lie – which makes Badinter's point for her, that the "essential mother" shtick is a way to crowbar women out of the marketplace. "Career-driven mothers" should be reminded, says Williams, "that family comes first."

But of course, feminist theorising on its own would be no match for the superiority of breastfeeding, if it were all it was cracked up to be. In government literature, both here and in the US, the benefits of breastfeeding are expressed uncritically, and to the farthest reaches of what's physically possible. The NHS Start4Life advice says that breastfeeding will give babies a smaller chance of, "developing eczema; getting ear, chest and tummy bugs and have [sic] to go to hospital as a result; being fussy about new foods; being constipated; being obese and developing diabetes when they are older. There are advantages for mums who breastfeed too: breastfeeding lowers the risk of breast and ovarian cancer; breastfeeding naturally uses up about 500 extra calories a day so mums who breastfeed often find it easier to lose their pregnancy weight."

Now, only one of those claims is without dispute, as Wolf concluded, the one about "tummy bugs". The research about ear infections, respiratory disease and diabetes is very mixed. Neither fussiness around new foods nor constipation are classic or very salient markers of good health. The obesity studies are debatable. The weight loss of breastfeeding mothers is taken from a WHO report, based on two interventionist studies in Honduras. It's nothing like that straightforward: it does use 500 calories a day, but if your appetite increases at the same time, it is amazing how quickly and easily you can pop 500 calories into your mouth. Meanwhile, Wolf points out, no study on maternal cancer has "distinguished the effects of breastfeeding from the behaviour of women who breastfeed". And that's the rub, as she writes: "Breastfeeding cannot be distinguished from the decision to breastfeed, which could represent a more comprehensive commitment to healthy living." It's a self-selecting sample, a phrase that is a foreign language in the world of early-years intervention.

The interesting thing is, as Wolf told me, these problems aren't in dispute: "The American Academy of Paediatrics released a statement on breastfeeding, with a point in the policy document where they say, 'There are a lot of methodological problems'. And then they continue as if those problems didn't exist. It's like saying something with your fingers behind your back – you can carry on with your business as though those problems weren't there."

Wolf also quotes Diane Eyer, a psychologist and academic, in this observation that has consequences right across the early years debate: "The notion of risk [has been] transformed from a dichotomy to a continuum." It's no longer a case of "safe" versus "dangerous"; rather, everything carries some risk and you announce your fitness as a parent to the world by interpreting hazards in the most credulous, fervent way. The onus isn't on the researcher to prove the point any more – the onus is on the parent, or parent-to-be, to prove that they'll believe the researcher. This leads to some utterly bogus propositions being taken as fact (don't get me started on stress in pregnancy).

At the same time, at the level of policy, there is a zealous belief in the importance of the first three years in fashioning good citizens, which you can see in everything from David Cameron's parenting classes to Frank Field's idea for a school in which the pupils are enrolled in utero (I'm not kidding). The assumption tends to be that the kind of people who breastfeed anyway and eat organic have no need of advice, while the people to whom advice is dispensed are essentially counselled to act more like the middle classes. And it makes sense on its own terms, since if you've decided that social mobility will be achieved by intervening at the parental level with the under-threes, you have basically decided this: if only everybody would parent (there's really no other verb for it) the way the richest parent, then everybody would have the same chances as currently only the richest enjoy. One precept this relies on is an amplified belief in the risks of certain types of parenting over other types; the minute you accept the idea of different but equal – formula feeding, say, as different from breastfeeding, but fine all the same – you abnegate your right as a government to stick your oar in.

Leaving social policy and returning to the world, it is interesting to note that the standard advice is to breastfeed exclusively ("don't rush to mush!") for six months, and yet only half of mothers in the UK who start breastfeeding (an impressive 91%) continue past six weeks. So there's this very trenchant public health message that the majority of people aren't taking any notice of – in that regard, it's a bit like the five-a-day message in adult nutrition (ie, we mainly ignore it), and yet it's unlike the proscription of drink-driving, which has pretty good social penetration. The message I take from that is that we tend to supplement government advice with the evidence of our own five senses, and only when the two are in accord do we take it all that seriously.

I didn't set out to write a controversial book about pregnancy. (Within the sappy remit of thinking your kids are the best thing that ever happened to the world, how controversial is it going to get, realistically?) And yet there are wells and eddies of the baby world that reflect things about class and gender and authority and control that are not entirely babyish.

• What Not To Expect When You're Expecting, by Zoe Williams, is published in paperback by Guardian Books on 7 June at £7.99. To order a copy for £5.99, including UK mainland p&p, go to guardianbookshop.co.uk or call 0330 333 6846.