NeuroTrailblazers banner

Career Mechanics

How the machinery of a research career actually works — funding, peer review, hiring, CVs, collaborations, and who pays for conferences — written for someone who has never seen any of it from the inside.

Before you start

   
Time ~45 minutes to read, then one afternoon of the actions
You need The name of one lab, program, or company you are actually near; a calendar you use
You finish with Five questions asked out loud that you were previously guessing the answers to
Assumes Nothing. This page assumes you have never seen a grant budget, a decision letter, or a search committee

Module 24 tells you how to choose a next step and how to write to the people who control it. This page is about the machine those steps run inside: where the money is, who decides, and on what timetable. The machine is not secret, but it is undocumented, and people who grew up near it learned it by overhearing.

The practical claim of this page: almost every question below can be answered by asking one person one sentence. The barrier is not access to the answer. It is not knowing that the question is askable, and not knowing the words. So each section ends with the words.

1. Where a salary comes from, and why your position is contingent

The structure. A funding agency or foundation makes an award to an institution, not to a person. A named principal investigator is responsible for the science and the budget. The award runs for a fixed term. Inside it, a budget line pays some fraction of specific salaries; another slice, the indirect or overhead rate, goes to the institution rather than to the project. Your stipend, wage, or salary is one of those lines.

Three consequences that change what you do:

Cycles. Agencies publish funding opportunities with submission deadlines, then run review panels, then an advisory step, then issue awards. The gap between “we submitted” and “money exists” is long enough that labs routinely bridge it. Do not take a timetable from this page — timetables differ by agency, mechanism, and year, and any date printed here would be wrong somewhere. Look up the specific funding opportunity announcement and the agency’s own posted review calendar, and ask the lab’s administrator what their submission history looks like.

Fellowships versus grants — the distinction with the most leverage in it. A grant is awarded against a PI’s project; a fellowship is awarded against you and generally moves with you. A funded trainee is cheaper to host, which is why a fellowship changes the answer to “do you have space” in a way that enthusiasm does not. If you are choosing what to spend a month on, an application that makes you portable usually beats one that does not.

Vocabulary you will hear and nobody will define. Soft money — a position paid entirely from grants rather than from an institutional line, so it ends when the grants do. Bridge funding — institutional money covering a gap between awards, usually discretionary and finite. No-cost extension — permission to spend remaining money past the original end date, which extends the position without adding money to it. Effort — the fraction of a person’s time formally charged to an award.

The script: asking what pays for you

Ask it when you accept a position, at your first one-to-one, or at the start of an academic year. Not in a crisis — in a crisis it reads as an accusation.

“I want to plan properly rather than assume. Which award is my position charged to, when does it end, and is a renewal or a follow-on in preparation? If the answer is uncertain, that’s useful too — I’d rather know the range.”

Each clause is doing work. “I want to plan properly” gives the question a purpose other than doubt. Naming award, end date, and renewal as three separate items means a vague answer is visibly vague. “If the answer is uncertain, that’s useful too” is the important one: it releases the PI from having to sound confident, which is the thing that otherwise produces a reassuring non-answer. Dr Nguyen, in the mentor persona, is holding exactly this uncertainty and is unlikely to volunteer it unprompted.

Who this section is for. Julian, in the first-generation persona, will not ask, because in every job he has held, asking about the employer’s finances would have been out of order. In research it is normal and the information is often public. Amir, the industry incomer, has the opposite error: he will assume a headcount model where a position is a slot that persists until someone cancels it. There is no slot. There is a budget line with an end date.

2. Peer review, end to end, and what an editor actually does

The sequence for a typical journal:

  1. Submission. The corresponding author uploads the manuscript, a cover letter, and often a list of suggested and excluded reviewers.
  2. Editorial triage. An editor decides whether it goes out for review at all. A rejection here is a desk rejection and is usually about scope, novelty framing, or obvious method gaps — it is fast and it is not a scientific verdict.
  3. Reviewer invitation. The editor invites reviewers. Many decline. This is the step that takes unpredictable amounts of time and the one authors most often misattribute to reviewers being slow.
  4. Reviews return as free-text comments to authors, plus confidential comments to the editor and a recommendation.
  5. Decision. The editor writes a decision letter: reject, major revision, minor revision, or accept. Straight acceptance on first pass is rare.
  6. Revision and response. Authors revise and write a point-by-point response. See Module 17.
  7. Re-review or editor-only decision, then acceptance, then production, then publication.

What the editor actually does, which is the part nobody explains. The editor is not a super-reviewer and generally does not re-derive your statistics. The editor decides scope fit, chooses who reviews, weighs conflicting reviews, and writes the decision. Reviews are advisory. The decision letter is the instruction.

That has a direct operational consequence: when two reviews conflict and you are deciding what to spend three months on, the editor’s own paragraph tells you which concerns are load-bearing. Read the decision letter first, then the reviews, then the decision letter again, and build your revision plan from the editor’s sentences. Junior authors routinely do the opposite — they respond exhaustively to the harshest reviewer and under-address the point the editor flagged — and then are surprised by a second round.

Appeals exist and are rarely used. An appeal is for a demonstrable error: a reviewer asserted something factually wrong about your method, or reviewed a version you did not submit. It is not for disagreement about importance. Editors will read a short, specific, unemotional appeal; they will not reopen a taste judgement.

Preprints change the timeline, not the process. Posting a preprint puts a dated, citable version on the record while review runs. Journal policies on preprints vary, so check the specific journal before posting rather than assuming.

The script: getting inside the process before you need it

Two moves, in order.

Move one — read one complete file. Ask a supervisor or any published author:

“Would you be willing to walk me through one paper’s full review file — the submission, the reviews, your response letter, and the decision? I have never seen what a decision letter looks like and I’d rather see one before I have to answer one.”

The reason to ask for the whole file rather than “advice on reviewing” is that the artifact teaches what description cannot: how blunt reviewers are, how ordinary a major revision is, and how a response letter is structured. Most people say yes, because it costs them nothing but a folder.

Move two — co-review, named. Reviewing is how you learn what reviewers actually check, and how editors learn your name. You get asked once you have published, or once someone passes you an invitation. Do not ghost-write a review under someone else’s name; ask to be named:

“If you take that review, could I co-review it with you and be named to the editor? I’d draft the methods section of the review, you’d take the rest and have the final say on everything I write.”

Naming matters because a named co-review is a record — it goes to the editor and it belongs on your CV — and an unnamed one is invisible labour. The scoped offer (“I’d draft the methods section”) makes it cheap to accept rather than a favour. Module 19 covers what to write once you are in the chair.

3. Hiring: academia and industry are different machines

  Academic search Industry loop
Trigger A funded line, advertised with a closing date A team’s need, posted continuously and filled when found
First filter A committee reading applications and reference letters together A recruiter screening against a requirement list
Central evidence Trajectory, letters from people who watched you work, a research vision Demonstrated skill in a bounded exercise, in the next two hours
The visit A seminar plus, often, a chalk talk on future work; meetings across a department A sequence of interviews in one or two days, plus a debrief meeting you never see
Decision speed Months, with long silences that are not signals Weeks, sometimes days
What is negotiable Start-up package, space, teaching load, start date — often more than salary Salary, level, equity, start date
Where referral acts On whether your letters are read closely and who is invited On whether your application is looked at at all

The difference that should change what you build. Academic hiring runs on people who can testify to your work over years. Industry hiring runs on your ability to show the skill inside a bounded exercise, to strangers, today. If you want both doors open — which the Module 24 decision table treats as a live option — you need both assets: three people who have watched you work, and a portfolio a non-neuroscientist can read without you in the room. Building either takes longer than a job cycle. Start now, not when the posting appears.

Two persona-specific corrections.

Amir arrives with an industry reflex: find the posting, apply through the portal. In academia that reflex fails twice over. Positions are frequently shaped around a person’s fit before the ad closes; contacting people before applying is normal and expected rather than pushy; and the portal is a compliance step, not the channel. The correction is in Module 24’s outreach material — use it, and use it months earlier than feels reasonable.

Julian has the mirror problem: he will over-value the application and under-value the letters, because the application is the visible artifact and the letters happen in a room he cannot see. Letters are frequently the highest-weight item in an academic file. A strong letter requires someone to have watched you make decisions over months. That is a two-year build, not a two-week request, and it is the strongest argument for taking a scoped research role early even when it is unglamorous.

Numbers, deliberately absent. Salary levels, offer rates, and time-to-hire vary by country, sector, institution, and year. Do not take them from a curriculum page, including this one. Get them from three sources: current job postings in the exact role, your institution’s published pay scales where they exist, and two people who currently hold that job and will tell you a number.

4. What a CV is for and how it is actually read

A CV is not a record of your life. It is a claim-evidence document, skimmed in somewhere between thirty and ninety seconds by a person whose immediate task is to make a large pile smaller. Everything follows from that reading behaviour.

What the reader looks at, roughly in order: where you are now and for how long; whether there is a first-author output and where it sits; whether the dates are continuous and, if not, whether the gap is explained; whether your named skills are matched by artifacts elsewhere in the document; and, near the end, whether anything is inflated, because one inflated line devalues the whole page.

Rules that are enforced silently.

The script: the consortium contribution line

Not a sentence you say — a sentence you write, and the one most often written badly.

Weak: “Contributor, whole-brain proofreading effort, 2025–2026.”

Strong: “Proofreading and QC, whole-brain reconstruction effort (2025–2026): triaged and corrected 340 segments in the layer 2/3 volume; wrote the deferral-rate QC check now used by the group. Segmentation and error flags produced by the consortium pipeline.”

The strong version names the verb, the count, a thing that outlived your involvement, and — critically — what you did not do. Volunteering the boundary is what makes the rest of the claim credible, which is the same mechanism as the evidence caption in Module 25.

For Julian specifically: put your paid employment on the CV. The instinct is to hide twenty hours a week of unrelated work because it looks like it competes with research. It does compete with research — that is precisely the point. A reader looking at a slower output record with no employment listed infers something worse than the truth. Listed, it is context. Hidden, it is an unexplained gap.

5. How collaborations form, and what one costs you

How they form. Almost never from a broadcast offer to collaborate. They form when one person states a specific unmet need out loud and another person recognises that they have the thing. This is the entire mechanism, and it is why Module 23’s advice to open with a concrete one-line project framing is not small talk — a specific opener is what makes recognition possible. “I’m looking at whether reciprocal E-I pairs are enriched in layer 2/3, and my bottleneck is a distance-matched null” gets a collaborator. “I’m interested in network motifs” does not.

What it costs. A collaboration is not free, and the costs are usually unpriced at the start:

The one-page memo, written before the work starts. Seven lines: the question; what each side provides; the data-use terms and what may be shown publicly; who does which task; the authorship expectation as currently understood and what would change it; what happens if the project stops; and who may publish what in that case. It takes twenty minutes and it is the single highest-return administrative act in this entire page.

The script: proposing the memo without sounding distrustful

The obstacle is social, not intellectual. Proposing terms feels like accusing someone of planning to cheat. Solve it by attaching the memo to scope, not to trust:

“Before I start, can we write half a page on scope — what I’m delivering, what you’re delivering, and roughly where this would sit for authorship as we currently see it? I mostly want it so that if it grows, we both notice we’re renegotiating rather than drifting.”

The moves: “before I start” places it at the only moment when it is cheap. “what I’m delivering” first makes it an offer rather than a demand. “as we currently see it” concedes that it may change, which is what makes it signable. “we both notice we’re renegotiating” names the actual failure mode — silent scope growth — and puts you and them on the same side of it. Almost nobody refuses this. A person who does refuse it has told you something useful for free.

The script: the ask you should decline, or resize

Maya, in the graduate persona, is the one this happens to. She is the bridge person, so people ask her for method help constantly, and the help arrives as a favour and departs as an acknowledgement. Repeated enough, it is a serious tax on her own first-author work.

“I can’t take on the full analysis this term. What I can do is a two-hour session where I walk your student through the pipeline and hand over my notebook, and then they own it. If you’d rather I ran the analysis myself, that’s a bigger piece of work and I’d want to talk about where it sits for authorship first.”

Three moves. The refusal is of the scope, not the person. A smaller concrete alternative is offered immediately, so the answer is not simply no. And the authorship question is attached to the larger option only, which makes raising it a natural consequence of size rather than an act of suspicion.

6. Conference economics: what it costs and who pays

Module 23 covers which venue to choose and what to do when you are there. This is the money.

The cost lines, all of which are separate and all of which surprise first-time attendees: registration, usually with a cheaper early rate and a trainee or student rate; society membership, which is sometimes a precondition for the trainee rate or for submitting an abstract; an abstract submission fee at some meetings; travel; lodging, where the conference hotel is convenient and rarely the cheapest; food across several days; visa application costs and appointments where relevant, with long lead times; and care costs at home while you are away.

Where the money comes from, in the order worth trying:

Do not take any deadline from this page. The rule that generalises: the money deadline is earlier than the abstract deadline, and both are earlier than you think. When you decide to target a meeting, spend one hour on its website finding the award deadlines and put them in the calendar before you write a word of the abstract.

The barrier nobody names: reimbursement. Many mechanisms pay you back after the trip. That means fronting a substantial amount for weeks. Whether that is an inconvenience or a hard stop depends entirely on whether someone can float it for you, which is a fact about your family, not about you. This is one of the clearest places where the machine hands an advantage to people who already have one, and it is frequently solvable — travel advances, direct institutional billing of registration and hotel, and pre-paid bookings by an administrator are all routine. They must be requested, usually weeks ahead, and nobody will offer.

The script: asking about advance payment

To the PI or, better, to the departmental administrator, who actually knows:

“I’d like to go to the meeting in [month] and I’ve found the trainee registration rate. Before I book — is reimbursement the only route, or can registration and the hotel be billed directly or advanced? I’d rather set it up now than carry it on a card.”

The moves: you have already done the work of finding the rate, so this is not an open-ended request. You ask about the mechanism, not for a favour, which makes it a process question an administrator answers all day. “I’d rather set it up now” gives a reason that requires no disclosure of your finances. You are entitled to say more than that if you want to, but nothing here requires you to.

7. Where this machinery is unfair, stated plainly

Not as consolation. As a list of places where knowing the mechanism converts an invisible disadvantage into an ordinary task.

The mechanism Who it quietly favours What closes the gap
Position funding is undiscussed People with a parent or family friend who has held a grant and explains the term structure The script in §1, asked once a year
Peer review is invisible until you are an author People whose supervisor showed them a decision letter as a trainee Ask to read one complete review file (§2)
Academic hiring runs on letters People placed early in labs where senior people watched them work Deliberately build three long-horizon referees; take a scoped research role early
Pre-application contact is expected but unwritten People told at home that you email supervisors first Module 24’s outreach scripts, sent a year early
Collaboration terms are informal People whose informal networks make reneging socially expensive for the other side The half-page memo (§5), proposed as scope
Conference funding is reimbursement-first People who can float several thousand for weeks Ask for advances and direct billing (§6), weeks ahead
Unpaid or under-scoped research posts People who do not need to work for money during term Refuse to treat unpaid as the only route; the Module 24 decision table has paid rows in it

The through-line: none of these are gaps in ability, and every one of them is closed by a question a person is entitled to ask. That is the whole argument for making the hidden curriculum explicit — it is the fairness intervention described in Education Models, not an add-on to the technical training.

Check yourself

Your PI says "we might not have funding for you next year, but don't worry about it." What do you do this week?

Convert the sentence into three facts and one date.

Ask, in writing so you both have the record: which award the position is charged to, its end date, and whether a renewal or new submission is in preparation and when its decision is expected. Then put the earliest of those dates in your calendar with a six-month lead reminder, because that is when your alternatives need to be live rather than hypothetical.

The instruction “don’t worry” is usually sincere and usually means the PI intends to solve it. It is not information you can plan against. Note that this is not distrust: it is the same practice as the stopping rule in Unit 08 — you decide in advance what state of the world triggers a decision, rather than deciding continuously under pressure.

Two reviews come back. Reviewer 1 is hostile and demands three new analyses. Reviewer 2 is mild and asks for a clarification. The editor's letter says "please address the concerns about reconstruction quality." Where does your effort go?

Into reconstruction quality first, whichever reviewer raised it — and if neither raised it in those words, the editor has told you what they themselves are unconvinced by, which is more important still.

The editor writes the decision. Reviewers advise. A response letter that exhaustively satisfies the hostile reviewer while giving the editor’s own sentence a paragraph is a common and expensive mistake, because the person deciding did not get their answer.

Do address all three of Reviewer 1’s requests — but the correct response to a request you decline is a reasoned decline, not silence, and the editor is the audience for that reasoning.

A senior researcher at another institution asks you to "run your pipeline on our data, it should be quick." What is the first thing you write?

Half a page of scope, before any code runs — §5’s memo, using §5’s script.

“Should be quick” is a scope estimate made by someone who has not used the pipeline. The predictable path is that it is not quick, that it grows, that it becomes the reason your own analysis slips a term, and that at submission you discover you are in the acknowledgements.

The memo does not require distrust, and proposing it does not damage the relationship — it is a scope conversation, framed as one. If the answer to “roughly where would this sit for authorship” is evasive, you have learned something at a cost of twenty minutes rather than at a cost of six months.

Common errors and how to recover

You have been in a position for a year and do not know when it ends. Ask this week, using §1’s script. If asking now feels late, note that it becomes more awkward every month, which is an argument for today rather than an argument against.

You applied through the portal and heard nothing. For academic positions, the portal is not the channel. Silence there is not a verdict. Recover by contacting people directly, months earlier in the next cycle, with the specificity Module 24 requires.

Your CV lists three papers “in preparation.” Replace with what is checkable. Post a preprint if the work supports one, or describe the work as a project with an artifact someone can open. One verifiable item outranks three aspirational ones, and a reader who notices the pattern discounts the rest of the page.

You did a large piece of a collaboration and are now an acknowledgement. Fix the next one at the start, not this one at the end. For this one, decide what you actually want — the authorship position, an accurate contribution statement, or a strong reference — and pursue one. The routes are on the conflict page.

The abstract was accepted and you have no money. Ask immediately about the trainee rate, remaining departmental funds, and whether registration can be billed directly; some awards have late or rolling rounds. Then, regardless of outcome, put next year’s award deadlines in the calendar today. This failure repeats annually until the calendar changes.

What this page does not cover