Research foundation
Research as a starting point.
Your behavior as ground truth.
Axis takes the published research seriously — and takes its limits seriously too. What follows is the actual literature behind the product, including where it’s well-replicated, where it’s only directional, and where the most-cited studies have failed to replicate. We start with what the research suggests and then learn who you actually are.
How to read this page
Each domain below carries a confidence label about the underlying science. We don’t pretend research that has failed to replicate is settled — and we don’t pull a feature out of the product when the citation that motivated it weakens. The product behavior earns its keep on its own.
Where the research is contested — or where it’s settled but shows large individual variation — Axis intentionally builds in flexibility instead of locking in a number. Soft defaults you can override. Settings instead of hard rules. Learning loops instead of fixed thresholds. Personal variance isn’t noise we’re trying to suppress; it’s the signal we’re trying to fit to.
Well-replicated, broad consensus, robust methodology.
Directionally supported, with caveats — small samples, partial replications, or estimated thresholds.
The behavioral effect is well-supported across multiple lines of evidence. The underlying mechanism — why it works — is unresolved. Axis implements the action regardless.
On this page
01
StrongContext switching & attention residue
Switching between tasks leaves a cognitive residue that degrades performance on the next task. Recovery isn’t instant — and the cost compounds across a day full of small switches.
What the research says
Attention residue from one task degrades performance on the next, with the largest losses observed when the prior task was incomplete.
Leroy, 2009
Caveat: The frequently quoted “40% performance loss” is a worst-case maximum under specific conditions, not a universal figure.
Recovery from a switch typically takes 15–25 minutes before full re-engagement.
Leroy, 2009
Complex task switches produce significant timing losses on the new task.
Rubinstein, Meyer & Evans, 2001
How Axis applies it
Axis encodes adjacency rules between modes. The engine refuses to sandwich a deep-work block between two meetings without recovery. RECOVERY blocks are auto-inserted after modes flagged as high-drain, and the AI surfaces a soft warning when a proposed schedule violates a switching-cost rule.
02
ModerateDeep work & flow states
Sustained, uninterrupted focus produces qualitatively different output than fragmented work. Specific time thresholds are convergent heuristics from multiple sources rather than measured limits.
What the research says
Flow requires extended uninterrupted time, often described as somewhere in a 60–120 minute range.
Csikszentmihalyi, 1990
Caveat: These are practitioner-reported heuristics, not measured ceilings. Individual variation is substantial.
Most people hit diminishing returns on deep work somewhere in the 3–6 hour daily range.
Ericsson et al., 1993; converging evidence from later work
Caveat: Ericsson’s original study has partially failed replication (Macnamara & Maitra, 2019). The heuristic still converges from multiple imperfect sources.
Different deep-work modes — convergent build, generative creative, analytical strategy — engage measurably different neural systems.
Beaty et al., 2016
Flow requires clear goals and immediate feedback, plus a balance between challenge and skill.
Csikszentmihalyi, 1990
How Axis applies it
Axis splits Deep Work into three distinct modes (Build, Creative, Strategy), each with its own ideal duration range. A combined daily deep-work limit (4–6h, user-adjustable) prevents the schedule from outrunning the research-suggested ceiling. Session feedback captures outcome and energy on every deep block — so over time the right ceiling gets learned, not assumed.
03
StrongChronobiology & circadian rhythms
Cognitive performance varies predictably across the day, but the timing of peaks varies by individual chronotype by several hours. Recent research (2024–25) emphasizes that alignment between schedule and chronotype matters more than any specific clock time being universally optimal.
What the research says
Individual chronotypes vary by 3–4 hours, distributed across a population.
Roenneberg et al., 2007
Schedule–chronotype alignment predicts work productivity in large samples.
Shimura et al., 2022 (n=8,155); KWSHS 2024 panel
A 2025 systematic review of 65 studies found consistent synchrony effects across cognitive domains.
Systematic review, 2025
How Axis applies it
Axis uses a six-option chronotype taxonomy (DAWN through NIGHT_OWL). The template generator places deep-work blocks around the user’s peak window rather than defaulting to mornings. The trends heatmap overlays chronotype on actual productive hours so misalignment becomes visible — and adjustable.
04
Mechanism debatedDecision load & pre-structured weeks
That sustained decision-making degrades downstream performance is well-supported across multiple paradigms — and that pre-designed structure protects capacity for the work that matters is one of the most robust practical findings in the literature. Why it works, on the other hand, is unresolved. The popular ego-depletion mechanism has failed replication. Axis implements the action without leaning on any single theory of why structure helps.
What the research says
Pre-designed structure removes hundreds of small “what should I do next” decisions across the day, preserving cognitive capacity for the work that matters.
Convergent across paradigms — implementation intentions (Gollwitzer, 1999), choice architecture (Thaler & Sunstein, 2008), pre-commitment (Bryan et al., 2010).
Process models — motivated disengagement, attention reallocation — fit the data on sustained decision-making better than the popular ego-depletion model.
Inzlicht & Schmeichel, 2012
The original ego-depletion model (decision-making drains a finite cognitive resource) has failed two large multi-lab replications.
Hagger et al., 2016; Vohs et al., 2021 (combined n > 5,600, effect size near zero)
Caveat: We acknowledge it because it’s the most popular framing — not because we depend on it. Pre-structured weeks reduce decision load whether the mechanism is depletion, motivation shift, or attention allocation.
How Axis applies it
Templates, archetype-seeded defaults, and the deterministic engine all reduce the same friction the literature is trying to explain. The user keeps full override on every block. Axis takes a position on the action — pre-structured weeks help — without taking a position on which mechanism is right.
05
StrongRecovery biology
Recovery isn’t a reward for working hard — it’s a biological requirement for the next round of high-quality work. Different kinds of recovery have measurably different cognitive value.
What the research says
Cognitive and physiological recovery is required for sustained performance, not optional.
Neuroscience consensus
Time in nature and movement-based recovery restore attentional capacity more effectively than passive rest.
Kaplan, 1989; Ratey, 2008
Passive digital consumption is low-quality recovery — it pulls attention without restoring it.
Research consensus
Burnout is staged, not binary — exhaustion → cynicism → inefficacy.
Maslach & Leiter, 2016
How Axis applies it
Recovery is a first-class domain in Axis, not a leftover. The engine auto-inserts recovery buffers after high-drain modes, and a NO_RECOVERY validation warning fires when a day stacks drain blocks without restoration. Recovery activities are differentiated by energy impact (a walk outside is not the same row as scrolling). The DRAIN mode (passive consumption) is detect-only — never planned — and shows up on trends so it stops being invisible.
06
StrongRituals & boundary management
Transitions between modes — into work, out of work, into the week, out of it — have measurable costs that can be reduced with structured rituals.
What the research says
A shutdown ritual that captures unfinished tasks reduces the cognitive interference that incomplete tasks create after hours.
Masicampo & Baumeister, 2011
Morning reattachment to work predicts daily work engagement.
Sonnentag & Kühnel, 2016
Boundary-control behaviors reduce work–life interference.
Ashforth et al., 2000; Clark, 2000
Pre-commitment devices are an effective lever for behavior consistency.
Bryan et al., 2010
How Axis applies it
Axis ships six ritual types — Morning, Startup, Shutdown, Evening, Week Startup, Week Shutdown — each with an ordered checklist of steps. Startup is framed as re-engaging with work before the day begins; Shutdown is framed as disengaging so recovery can actually start. Rituals are auto-scheduled into the week so they don’t depend on memory.
07
StrongBehavioral architecture
How goals are set, structured, and connected to time has predictable effects on follow-through. This domain is where research and product alignment is tightest.
What the research says
Specific, difficult goals consistently produce better performance than vague or easy ones.
Locke & Latham, 1990 (and 30+ years of replication)
Implementation intentions — “when X happens, I will do Y” — substantially raise follow-through over generic intent.
Gollwitzer, 1999
Autonomy, competence, and relatedness are the conditions for sustained intrinsic motivation.
Deci & Ryan (Self-Determination Theory)
Resource conservation theory predicts how energy depletion affects later decision quality.
Hobfoll, 1989
How Axis applies it
Goals in Axis carry a why, an estimated effort, a priority, and (optionally) milestones. Tasks are bound to goals; blocks are bound to goals; hours roll up automatically. The advisor flags goals where intent and behavior have drifted apart. Nothing in the system enforces, scolds, or punishes — autonomy is the architectural default, because the research says it’s how follow-through actually compounds.
A note on honesty
On research we don’t cite.
You won’t find the “35,000 decisions a day” figure anywhere on this site or in the product. It has no credible primary source. Other popular numbers — the exact minutes for ultradian recovery, fixed thresholds for “how much deep work a day” — get qualified as heuristics rather than measured ceilings.
We publish this because productivity tools that lean on retracted research while claiming to be science-driven are a real category — and we don’t want to be in it. The product is built so no single citation is load-bearing. Research is the starting point. Your behavior is what gets weighted over time.
Sources: Implementation Mapping v2.0 (April 2026), Synthesized Research Review v1.0, and Product Spec v3.12. Citation list and full claim table available on request — info@axisplanner.ai.
See how it actually works.
The science is upstream of the product. The product is what you’ll use.