books/theory/EVIDENCE.md

# Evidence tiers

*Every load-bearing claim in this project, sorted by how much weight it can actually carry. This file governs the register each chapter may be written in. If a claim isn't in here, it hasn't been checked yet — don't build on it.*

The rule: **a chapter can only be written in the voice its evidence tier permits.** Tier A can be stated. Tier B must present the dispute. Tier C must be marked as the author's hypothesis, in the text, not just in a footnote. Tier D is memoir or metaphor and cannot carry an argument.

---

## Tier A — Well-supported. State it, cite it, move on.

**The brain conserves energy by reusing models rather than building them.**
Predictive processing / free energy principle (Friston; Andy Clark), cognitive miser theory (Fiske & Taylor, 1984), System 1 / System 2 (Kahneman). Convergent from three independent traditions. The claim that heuristics and stereotyping are thrift rather than laziness is defensible as written.

**DMN and TPN exist, and are broadly anticorrelated.**
Standard, replicated resting-state neuroimaging. Safe.

**The salience network exists and is implicated in switching between them.**
Menon's triple-network model (2011); causal support from dynamic causal modelling and Granger analyses (Sridharan et al. 2008 and replications), with the right anterior insula governing executive-to-default transitions. *Caveat to keep visible:* "switchboard" is a functional simplification. The networks overlap, co-activate, and are not discrete modules. Say this once, early, and you've earned the metaphor for the rest of the book.

**Nature exposure reduces rumination.**
Bratman et al. 2015 (PNAS): 90 minutes in a natural setting reduced self-reported rumination *and* subgenual prefrontal cortex activation; 90 minutes in an urban setting did neither. This is the single best-evidenced practical claim in the whole project — it's a controlled comparison with both a self-report and a neural measure, and the urban walk is the right control. Build the nature chapter on this.

**Chronic stress raises appetite for energy-dense food.**
Adam & Epel (2007) and the surrounding cortisol-reactivity literature. Acute stress suppresses appetite; *chronic* elevation increases hunger and shifts preference toward sugar and fat. Note the acute/chronic distinction — it's the part popular writing gets wrong.

**Seasonal variation in brain response to cognitive tasks is real.**
Meyer et al. 2016 (PNAS): attention-related activity peaked near the summer solstice, working-memory activity near the autumn equinox. **Read this carefully — it supports "the brain varies across the year" and actively fails to support the DMN-winter / TPN-summer mapping.** Cite it honestly or not at all.

**Segmented sleep in pre-industrial Western Europe.**
Ekirch's historical documentation is solid *for that time and place*. See Tier B for the overreach.

---

## Tier B — Contested. You must present the dispute, in the text.

These are the ones that will get the book attacked if handled as settled. Handled openly, several of them become more interesting than the tidy version would have been.

**Biphasic sleep as a universal human pattern — no.**
Ekirch documents it for pre-industrial Europe. Yetish et al. 2015 studied three equatorial pre-industrial societies without electric light and found no regular extended night waking, concluding the pattern probably post-dates migration into Western Europe. Ekirch has responded that he never claimed universality. **The honest version: segmented sleep is a documented Western European historical pattern, not a human default.** That's still a good chapter — it just becomes a story about cultural variation rather than about recovering our nature.

**Seasonality of mood — much weaker than folk belief.**
Traffanstedt, Mehta & LoBello (2016) found no relationship between depression severity and season, latitude, or sunlight exposure in a large US survey sample, and argued the seasonal modifier lacks validity. Young (2017) published a rebuttal. This is genuinely unresolved. **This is the most dangerous single fact for Book Two** and must be met head-on in its opening chapters rather than avoided — the book's argument has to become "seasonal structure is worth designing" rather than "your mood follows the seasons."

**Dunbar's number.**
Lindenfors et al. 2021 re-ran the underlying regression with modern methods and got estimates from 16 to 109 with 95% confidence intervals spanning 2–520 — meaning no specific number is derivable. **Use the idea of a bounded social world; never use the figure 150.**

**Media multitasking impairs attention — small and inconsistent.**
Ophir et al. 2009 is the famous study; it largely failed to replicate. Wiradhany & Nieuwenstein (2017) ran two replications and a meta-analysis of the subsequent literature, finding an effect in the same direction but much smaller — around d = 0.17 — with a substantial minority of individual results pointing the opposite way. The effect may be real but is nothing like the size popular writing implies. Don't lean on it. *(Check the exact study count against the paper before citing a number in print.)*

**Digital detox — and this one helps you.**
Review evidence is poor quality (roughly 96% of studies rated weak, small non-representative samples). Total abstinence has the *weakest* record of any approach; some studies find no change and several find harm during the detox. Social-media-specific breaks do better, and therapy-style interventions that change *how* people use platforms outperform both limiting and abstaining. **This is direct support for the book's central practical claim: retriage beats abstention.** Use it.

---

## Tier C — Plausible, untested. Mark as hypothesis, in the text.

This is where the original contribution lives. Untested is not the same as weak — but it has to be labelled, every time, or the book is overclaiming.

- **The three-tier relevance model** (day's walk / day's ride / beyond the horizon). Nobody has tested this. It's the best idea in the project and it is entirely yours and entirely unvalidated. Say both things.
- **Tier collapse as the mechanism of information fatigue** — that distant information carrying local-urgency signals is what exhausts people, rather than volume or reward-hacking.
- **Salience-network flexibility as the lever**, rather than favouring reflection or focus. The theory's central practical claim.
- **The actionability gap** — that the specific harm of modern news is caring intensely about things you cannot affect. Currently the most under-developed idea in the project and possibly the most valuable.
- **Salience environment over step count** — that 5,000 steps somewhere quiet beats 10,000 in a corridor. Bratman gives partial support (setting mattered, distance didn't) but the strong version is untested.
- **Work-life balance complaints as switching failures rather than time-allocation failures.**
- **Desk hunger as cognitive load misread as fuel demand.** The cortisol literature (Tier A) supports stress→appetite; your specific interoceptive-suppression mechanism is your own and unverified.

---

## Tier D — Speculative or personal. Cut, or flag explicitly as speculation.

**Cut entirely:**

- **The foreign-policy claim** (TPN-dominant societies bias toward interventionism). Unfalsifiable, politically loaded, and a single quotation from it discredits everything else in the book. Nothing is lost by removing it.
- **The DMN-based weight-loss protocol.** A prescriptive four-anchor health regimen resting on an invented mechanism. "Weight regulation follows nervous-system safety, not willpower" is an unsupported clinical claim about a population that includes vulnerable readers. The underlying *observation* — hunger after desk work, none after trekking — survives as a personal essay. The protocol does not.
- **The diagnostic test's numeric scoring.** Thresholds like "≥12 indicates imbalance" confer clinical precision derived from a single respondent. Keep the twenty questions as reflection prompts with no scoring, no cutoffs, and no diagnostic language.
- **The "I" / "me" labels for DMN and TPN.** These invert what most readers intuit, add no explanatory power, and create an unnecessary defence burden. Drop them.

**Demote to explicit metaphor:**

- **The seasonal DMN/TPN/SN mapping.** No evidence supports the specific assignment, and Meyer 2016 points the other way. It survives — usefully — as a *design metaphor* for structuring a year, which is how Book Two should frame it from page one.

---

## The internal contradiction to resolve before drafting

The current theory treats DMN activity as the thing modern life starves us of, and therefore as good. But:

- Killingsworth & Gilbert (2010) found mind-wandering — the DMN's signature activity — predicted *lower* momentary happiness, and accounted for more variance in happiness than what people were actually doing.
- Rumination, which the theory correctly lists as the DMN's failure mode, is DMN-linked.
- Bratman's nature finding is a study of nature *reducing* self-referential neural activity.

So "we don't get enough DMN" and "nature is restorative because it quiets self-referential thought" are pulling against each other. The resolution the theory already contains — that the target is *flexibility*, not more of either mode — is the right one, but it has to be stated explicitly and early rather than left implicit. **This is the argument chapter 7 of Book One exists to make.** Handled well, it's a strength: it's the point where the book stops being another "reflect more, scroll less" book.

---

## Sources

- [Bratman et al. 2015, *PNAS* — Nature experience reduces rumination and subgenual prefrontal cortex activation](https://www.pnas.org/doi/10.1073/pnas.1510459112)
- [Yetish et al. 2015 — Natural sleep and its seasonal variations in three pre-industrial societies](https://gurven.anth.ucsb.edu/sites/secure.lsit.ucsb.edu.anth.d7_gurven/files/sitefiles/papers/yetishetal2015.pdf)
- [Ekirch 2016, *SLEEP* — Segmented sleep in preindustrial societies](https://academic.oup.com/sleep/article/39/3/715/2454050)
- [Traffanstedt, Mehta & LoBello 2016 — Major depression with seasonal variation: is it a valid construct?](https://journals.sagepub.com/doi/abs/10.1177/2167702615615867)
- [Young 2017 — Does seasonal affective disorder exist? A commentary](https://journals.sagepub.com/doi/abs/10.1177/2167702616689086)
- [Lindenfors, Wartel & Lind 2021, *Biology Letters* — 'Dunbar's number' deconstructed](https://royalsocietypublishing.org/doi/10.1098/rsbl.2021.0158)
- [Wiradhany & Nieuwenstein 2017 — Cognitive control in media multitaskers: two replication studies and a meta-analysis](https://pubmed.ncbi.nlm.nih.gov/28840547/)
- [Digital detox and well-being — review of evidence quality](https://pmc.ncbi.nlm.nih.gov/articles/PMC11422191/)
- [Social media detox and well-being: a meta-analysis of RCTs](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11939267/)
- [Killingsworth & Gilbert 2010, *Science* — A wandering mind is an unhappy mind](https://www.science.org/doi/10.1126/science.1192439)
- [Adam & Epel 2007 — Stress, eating and the reward system](https://www.sciencedirect.com/science/article/abs/pii/S0031938407001278)
- [Systematic review of fMRI studies on seasonal modulation of brain functional connectivity](https://www.imrpress.com/journal/BJHM/87/5/10.31083/BJHM53980)
- [Triple network model, dynamically revisited](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7027374/)