Review of Johann Hari’s book Lost Connections

by Jacob Williams on 2025-02-12 PDF
feedback is welcome at jacob@brokensandals.net

I’d summarize some of this book’s main claims as follows:

  1. Healthcare systems treat depression as purely a brain disorder. When people are depressed, doctors ignore their life circumstances and assume they need antidepressant medications.
  2. Antidepressant medications don’t work very well.
  3. External factors play a huge role in depression.
  4. Changes to our lifestyles and our societies can dramatically alleviate or cure depression.

As other reviewers note, #1 seems like an exaggeration. Anecdotally, the nurse practitioner who first prescribed antidepressants to me also asked about what was going on in my life (spoiler: a breakup). My impression is that the strain of thought Hari is fighting against does exist in society to some degree though—there are some people out there who assume their unhappiness must be the result of some internal defect when it could really be adequately explained by how much their lives suck.

Hari hedges on #2 a bit; he notes that “some credible scientists argue [chemical antidepressants] give some temporary relief to a minority of users, and that shouldn’t be dismissed.”1 This comes across as a grudging admission, though. I worry this book may scare people away from antidepressants who could benefit from them. I do think that for many—most?—of us, antidepressants should only be part of a larger plan for dealing with depression, not the whole plan. But when you’re facing an overwhelming problem it’s wise to use every tool at your disposal in fighting it. When I was very suicidal, I don’t think any individual thing I did—meds, reaching out to friends, making radical life changes—was enough to get me through, but all of them together did. And the meds were useful even though I believed from the outset that my issue was fundamentally about circumstances, not brain chemistry. (I do think it’s possible that the meds only benefited me via placebo effect; like Hari, I felt like they kicked in faster than I was told they would.)

Hari also admits #3 may not be true for everybody. He says he “asked many people involved in treating depressed patients if they believe endogenous depression—the kind caused just by a malfunctioning brain or body—is real. They disagreed with one another. … But everyone agreed that if it exists at all, it’s a tiny minority of depressed people.”2 Just before calling it “a tiny minority” though, he says one of the doctors he asked “told me it could be as many as one in twenty of the people who come to him with depression.”3 I would argue 5% isn’t “tiny”, but whatever. I’m at least on board with the idea that many people are primarily depressed because of problems in our lives rather than problems in our heads.

Some of the solutions Hari puts forward will seem unsurprising or even cliché, like volunteering or doing loving-kindness meditation or (possibly and only with supervision) psychedelics. Others may sound hard to pull off (leaving your job to join or start a co-op) or like borderline pipe dreams (establishing a universal basic income)—although if Hari’s right about the transformative power of community, then perhaps just joining a group of people passionately trying to achieve such things can benefit you even if the goal is never accomplished.

I didn’t hear about Hari’s past credibility issues until I was already reading this book; otherwise, I probably would have been less willing to read it. Regardless, the important empirical claims in the book—e.g. about medication, meditation, psychedelics, even UBI—are the sort of things you’d never want to take just one book’s word for anyway, and I haven’t researched any of them enough to confidently hold opinions on them.

My notes on each chapter are available at https://brokensandals.net/notes/2025/lost-connections/


  1. Johann Hari, Lost Connections: Uncovering the Real Causes of Depression– and the Unexpected Solutions (New York: Bloomsbury, 2018), 312.↩︎

  2. Ibid., 183.↩︎

  3. Ibid.↩︎