Module 23: Posters, Abstracts, and Conferences

Prepare conference-ready abstracts and posters, and navigate scientific meetings with explicit hidden-curriculum support.

Use this module

Capability target

Submit a conference-ready abstract, produce a coherent poster draft, and deliver a defensible 90-second pitch with clear claim boundaries. Operationally: your abstract spends its word budget on evidence rather than motivation, a stranger can find your core result on the poster within ten seconds from two meters away, and you leave the session with at least two named follow-ups you initiated yourself.

Why this module matters

Conference communication is often the first high-stakes visibility point for trainees, and it is the point where the hidden curriculum bites hardest. The scientific content of a poster session is public; the operating rules are not. How to approach a poster where someone is already talking, how long a first conversation should last, whether it is acceptable to interrupt, what “send me an email” means and whether it means anything, when a business card is expected and when it is an affectation, how to leave a conversation without insulting anyone. Learners with academic families or well-connected undergraduate labs arrive knowing these by observation. Everyone else improvises, and improvisation under social pressure usually produces silence.

This module also sits on the MERIT stage-5 failure mode. If every professional relationship routes through your supervisor, you have borrowed a network rather than built one, and a conference is the cheapest place to fix that. The difference between a trainee who was introduced to eight people and one who initiated three conversations is invisible during the meeting. It is visible two years later, when one of them can email a stranger about a postdoc and the other cannot.

Concept set

1) Abstracts are claim filters

• Technical: every sentence should move from question to method to result to implication, with one explicit limitation. Budget the words before writing: roughly 25 for the gap, 20 for the question, 50 for dataset and method, 90 for results with quantitative anchors, 25 for the limitation, 20 for the implication. Anything that does not fit is being funded by the results sentences, which is the wrong trade.
• Plain language: include only what helps reviewers evaluate credibility and significance. A reader deciding whether to walk to your poster needs to know what you found, not what the field hopes for.
• Misconception guardrail: a strong motivating paragraph can compensate for thin results.

2) Posters are spatial arguments

• Technical: layout should mirror inferential flow (question, method, evidence, limits). Design for three reading distances: title and one-line finding readable from about two meters, headings and figure panels from a meter, methods and caveats only on close approach. Conventional minimums are a title around 72 to 100 point, headings 36 to 48, body no smaller than 24; check your meeting’s template first, since some societies specify sizes.
• Plain language: if viewers cannot find your core result in 10 seconds, redesign.
• Misconception guardrail: dense text signals that the work is thorough.

3) Conference performance includes social protocol

• Technical: introductions, question handling, and follow-up contact are part of dissemination competency and can be scripted in advance. The unit of success is not the conversation, it is the follow-up: a conversation with no agreed next action has produced enjoyment, not a network.
• Plain language: what you say and how you engage both matter, and both can be rehearsed.
• Misconception guardrail: good science speaks for itself and networking is optional decoration.

4) Key connectomics conferences and venues

Trainees should know the major venues where connectomics work is presented and discussed. The Society for Neuroscience (SfN) annual meeting is the largest general neuroscience conference and includes dedicated connectomics poster sessions and nanosymposia. The BRAIN Initiative Investigators Meeting brings together funded researchers and is a key venue for dataset announcements and tool demonstrations. Specialized connectomics conferences and workshops (often organized around specific datasets or methods) provide focused peer interaction. Gordon Research Conferences on neural circuits offer small-group, discussion-intensive formats that are especially valuable for early-career researchers building collaborations.

When writing abstracts for connectomics submissions, the 250-word limit demands precision. State the dataset used (species, brain region, reconstruction method), the analytical method applied, the primary finding with a quantitative anchor, and at least one interpretation limitation. Reviewers at these venues expect to see the specific dataset and version, not just “a connectomics dataset,” and they expect claims to be bounded by the reconstruction completeness and proofreading status of the data.

5) The follow-up is the product

• Technical: end every substantive conversation with one concrete option: a paper you will send, a notebook you will share, a question you will email. Write the person’s name and the agreed action down before the next conversation starts, because by the third day you will not remember. Send within 72 hours, while the memory of your poster is still attached to your name.
• Plain language: the value of a conference is realized in the week after it, or not at all.
• Misconception guardrail: if the conversation went well, the other person will follow up.

Hidden curriculum scaffold

The rules below are enforced consistently and stated nowhere. Give trainees the scripts, not the principles.

• Unspoken conference rules trainees often miss:
  • Start conversations with one-line project framing, not full backstory. “I’m looking at whether reciprocal excitatory-inhibitory pairs are enriched in layer 2/3” is a complete opener.
  • Answer questions to the level asked; escalate depth only if invited.
  • End interactions with a concrete follow-up option.
  • Approaching an occupied poster: stand at the edge of the group, listen, and wait for the presenter to finish an answer. If you have to leave before your turn, say “I’ll come back at half past” and do.
  • A senior person saying “send me an email” usually means it. Send it that week with a one-line reminder of who you are and what you discussed, because they had forty conversations.
• Inclusive support practices:
  • Provide sample scripts for first-time networking, including the exit script, which is the part trainees most often lack. A workable exit: “This has been really useful. I’ll let you get to the rest of the session. Can I send you the notebook?”
  • Normalize “I need to check that detail” responses. Model one in front of the group.
  • Debrief difficult interactions the same day. A trainee who was talked over will usually conclude the problem was them unless someone says otherwise.
  • Name the money. Registration and travel are usually the largest single expense a trainee meets in a year. Trainee rates, society travel awards, and institutional funds exist, and their deadlines typically fall months before the meeting. Put them in the calendar at the start of the planning cycle, not when the abstract is accepted.

Core workflow: abstract-to-poster pipeline

1. Distill the project into one claim hierarchy: one headline claim, at most two supporting claims, one limitation attached to each.
2. Draft the 250-word abstract against the sentence budget, then delete every sentence that does not help a reviewer judge credibility or significance.
3. Check the abstract names species, region, imaging modality, segmentation pipeline, version or materialization, proofreading state, and n. If any is missing, a reviewer will assume the worst case.
4. Build the poster wireframe around three evidence panels, placing the headline finding in the upper third where it is readable at two meters.
5. Write and rehearse the 90-second and 3-minute versions. The 90-second version is not the 3-minute version spoken quickly; it is a different script with the background removed.
6. Prepare a follow-up template and a target list of five to eight people you want to talk to, with one specific question for each.
7. Run critique for clarity, limitations, and the professional interaction plan, and revise before the submission deadline rather than after acceptance.

Choosing a venue: decision table

Trainees are usually told to “go to conferences” without being told which. The choice has real consequences and real costs.

Venue type	What you actually get	Who is in the room	How selection works	What it costs
Large general meeting (for example SfN)	Breadth, exposure to adjacent fields, many posters in your area	Thousands, including most people who might hire you	Poster acceptance is typically routine once membership and deadline requirements are met; talks are more selective	The highest registration and travel cost, and the highest risk of being one poster among thousands unless you plan your targets
Consortium or initiative meeting (for example a BRAIN Initiative investigators meeting)	Dataset and tool announcements before publication; the people who build the infrastructure you use	Funded groups and their trainees	Often tied to funding or invitation	Access may depend on your lab’s funding, a constraint worth naming rather than leaving trainees to infer
Small discussion conference (Gordon Research Conference format)	Sustained contact with senior people; questions that change your project	Roughly a hundred, same people all week	Application and selection, capacity-limited	Highest contact value per attendee, but unpublished material is typically off the record and not citable, so it will not generate a citable output
Computational meeting with competitive abstract review (for example COSYNE)	A methods-literate audience that will engage with your null model	Computational neuroscientists, many from machine learning	Competitive review; rejection is normal and not a verdict	Preparation cost is high and acceptance is uncertain, so do not make it your only submission
Local or regional symposium	Low-stakes rehearsal; local collaborators; a real deadline	Your institution and neighbors	Usually open	Least prestige, least cost, and the best place to fail safely before a national meeting

If you have never presented, a regional symposium followed by one large general meeting beats a single prestigious venue: the first converts the second from an ordeal into a repetition.

60-minute tutorial run-of-show

1. **00:00-08:00

   Framing + exemplar abstract**

   • Display a motivation-heavy abstract and its rewritten version side by side. Ask the group to count the words spent on the field versus on the result.

2. **08:00-20:00

   Abstract sprint (individual)**

   • Learners draft against the sentence budget with a visible word counter. Circulate asking only: “which sentence carries the number?”

3. **20:00-32:00

   Poster wireframe build (pairs)**

   • Wireframe on paper at one-eighth scale. Partners stand two meters back and report what they can read. Anything unreadable moves up or grows.

4. **32:00-44:00

   90-second pitch practice**

   • Deliver to a partner against a timer, twice, with the second delivery required to be shorter. Ban the background sentence on the second pass.

5. **44:00-54:00

   Q&A and networking simulation**

   • Rehearse the approach, one question, and the exit script. Rotate so everyone plays the senior person once; it is the fastest way to see how little a stranger retains.

6. **54:00-60:00

   Revision checklist + submission plan**

   • Each learner writes the submission deadline, the travel-award deadline, and two named people they intend to talk to.

Worked example: an abstract that spends its budget badly, then well

Julian’s first draft opens like this:

“The brain is the most complex structure known to science, and understanding how neurons connect is one of the central challenges of modern neuroscience. Recent advances in electron microscopy have made it possible to reconstruct neural circuits at synaptic resolution, opening unprecedented opportunities. Here we analyze connectivity in a cortical dataset and find interesting patterns of connectivity between cell types.”

That is 68 words of a 250-word budget, and it contains no dataset, no number, and no claim. The first two sentences are true of every abstract in the session, which means they distinguish nothing. The third sentence promises “interesting patterns”, which a reviewer reads as “the author has not decided what the result is”.

The rewrite reallocates the same budget:

“Cortical circuit models assume reciprocal connections between excitatory and inhibitory neurons, but reciprocity has not been counted at synaptic resolution. We asked whether reciprocal excitatory-inhibitory pairs in layer 2/3 are enriched relative to chance. Using a serial-section EM volume of mouse visual cortex, we extracted 1,247 reciprocal pairs among 12,891 possible excitatory-inhibitory pairs from a pinned materialization of the automatic segmentation, with synapses filtered at a fixed cleft-score threshold. Reciprocity was enriched 2.1-fold over a degree-preserving null model. The enrichment persisted in the manually proofread subset, though with a wider interval. False merges between excitatory and inhibitory arbors would inflate this measure in the same direction, and we have not yet quantified that contribution by merge injection. If the effect survives, it constrains the connectivity priors used in cortical circuit models.”

Three things changed. The gap replaced the field: “models assume X, nobody has counted X” is a gap; “the brain is complex” is not. Every claim acquired an anchor a reviewer can check, including the null model, because an enrichment without a stated null is not a result. And the limitation became specific and directional: saying “false merges would inflate this in the same direction” tells a reviewer you know which way your error pushes, which is worth more than a generic note that reconstruction has limitations.

The 90-second pitch is then a compression of the same structure, not a summary of the poster: one sentence of gap, one of question, one of dataset with version, two of result with the number, one of limitation, one invitation. About 180 spoken words.

The follow-up, sent within three days:

“Dear Dr Okafor — we spoke at my poster on Tuesday about reciprocal E-I connectivity in layer 2/3; you asked whether the enrichment survives merge injection. I’ve attached the notebook with the proofread-subset comparison, and I’m setting up the injection test this month. If it would be useful I can send the result when it runs. Thank you for the suggestion about the degree-and-distance null; I hadn’t considered the spatial constraint.”

Note the moves: it names when and where so she can place him, restates her own question so she recognizes it, sends something concrete rather than asking for something, and offers future contact without demanding a reply. It does not ask for a position; asking now would convert a professional contact into an awkward one.

Studio activity: conference simulation lab

Scenario: You have one accepted poster slot and must communicate your connectomics result to mixed audience members. Two of the people who stop are specialists who will ask about segmentation quality; one is a program director who will ask what it means; one is a peer at your own stage who wants your code.

Tasks

1. Produce one abstract against the sentence budget and one poster wireframe tested at two meters.
2. Deliver a 90-second pitch to a timer, twice, cutting on the second pass.
3. Respond to two audience questions, including one about reconstruction quality.
4. Draft one follow-up outreach message that sends something rather than asking for something.

Expected outputs

• Abstract draft.
• Poster wireframe.
• Pitch script.
• Follow-up message template.
• A named target list of five people and one question for each.

Assessment rubric

• Minimum pass: abstract claims are evidence-grounded and name dataset, version, and n; poster layout surfaces core result and limitation; the pitch communicates method, result, and limit within time.
• Strong performance: handles Q&A with precise uncertainty language, adapts to the audience in front of them without changing the evidentiary standard, and produces follow-ups that are specific, sent within days, and initiated rather than inherited from a supervisor.
• Common failure to flag: overstuffed abstracts with missing quantitative anchors, posters organized by chronology instead of argument, and overconfident responses to unresolved technical questions. Also flag a learner whose entire target list consists of people their supervisor already knows.

Common errors and how to recover

• The abstract deadline is in two days and the analysis is not final. Submit the claim you can defend now with the smaller n, not the claim you hope to have. Abstracts are routinely updated between submission and presentation; a retracted claim on a poster is far more expensive than a modest one in an abstract.
• Nobody stopped at your poster for the first thirty minutes. Stand to the side rather than in front of it, keep the headline finding at eye level, and make eye contact with people reading the title. If the session is genuinely dead, walk to two posters you flagged in advance and start there; you were going to do it anyway.
• You froze when a senior person asked a question. Use the stall that is not a stall: “Let me make sure I understand the question.” Repeating the question back buys five seconds and often clarifies what is being asked. Afterwards, write the question down; it is almost certainly one of the two you should have rehearsed.
• You collected contacts but sent nothing. Send late rather than not at all, and say so plainly: “We spoke at SfN about X; apologies for the slow follow-up.” A late message with a specific reference still works. A perfect message never sent does not.
• Every useful contact came through your supervisor. This is the stage-5 failure mode, and it is fixable only by deliberate practice. Before the next meeting, pick three people you will approach without an introduction, and ask your supervisor to be absent rather than present for those conversations.

What this module does not cover

• Building the talk itself. Slide design, the claim tree, and Q&A taxonomy are Module 22.
• The written paper. Manuscript structure, methods provenance, and reviewer responses are Module 17.
• Career decisions that conference contacts lead to. Program and role selection, outreach to prospective supervisors, and application timing are Module 24.
• The analysis behind the abstract. Null models and multiplicity are Module 20; this module assumes the result is already defensible.
• Visa, accessibility, and caring arrangements around travel. These are real constraints on who can attend which meeting and are handled by institutions rather than here. Raise them early with your program contact; the lead times are long.

Teaching resources

• Writing context: Module 17
• Presentation context: Module 22
• Evidence set: Journal Club
• Mentorship support: Ask an Expert
• Stage framing: Education Models — the stage-5 network-dependence failure mode
• Differentiation guidance: Facilitator Guide

Quick practice prompt

Write a 5-sentence mini-abstract including question, method with dataset version, result with one number, limitation, and implication. Then count how many words went to the field and how many to your result; if the first number is larger, rewrite the opening sentence.

Practice and progress

Check your understanding, then record your progress.

Abstract and Poster Checkpoint

Q1. Your 250-word abstract spends 90 words motivating the field. What does that cost you?

Nothing, if strong motivation is what makes reviewers score an abstract highly The results sentences, which are the words that let a reader judge credibility and decide to visit the poster The limitation sentence, which can move to the poster instead

The word budget is fixed, so every sentence is funded by another, and motivation is funded out of results. Roughly 90 words belong to results with quantitative anchors and about 25 to the limitation. Someone deciding whether to walk across the hall needs to know what you found, not what the field hopes for.

Q2. A colleague says your poster is fine because everything is readable when you stand at it. What test does the module actually set?

Title and one-line finding legible from about two metres, so a passer-by finds the result in ten seconds All body text at 24 point or larger, per the conventional minimum Figure panels sized so captions fit beneath them without wrapping

A poster is read at three distances, and the two-metre pass decides whether anyone stops at all. Point-size minimums serve that goal rather than replacing it: a poster can satisfy every typographic convention and still bury its finding mid-column. Design for the reader who has not yet decided to approach.

Q3. A senior researcher says "send me an email" after a good poster conversation. What does that mean?

It is a polite closing formula; act only if they follow up first Email them after the meeting, once you have something substantial to report They mean it: send within about 72 hours with a one-line reminder of who you are and what you discussed

They had forty conversations that week, so the reminder is doing real work and delay is what makes the message fail. Waiting until you have something substantial means arriving after your name has detached from your poster. This is the hidden-curriculum item with the widest gap between insiders and everyone else, and the conference's value is realised in the week after it or not at all.

Submit Quiz

Poster Q&A Microtask

You have 90 seconds and a specialist has asked about reconstruction quality. What goes in the answer?

That the dataset is a widely used public release, so its quality is established The proofreading state of the relevant arbors, the materialization version, and what those limit the claim to A promise to send the full quality-control report by email

Check Decision

Activity progress (self-reported)

State is saved locally in your browser for this module.

• Read capability target
• Complete studio activity
• Review assessment rubric
• Pass checkpoint quiz
• Complete microtask

0% of self-reported activities checked

Checkpoint: not yet passed.

Scored items update from your answers, not from checkboxes. Compare your work with the lesson rubric: this tracker does not certify proficiency.

Reset Progress

Teaching Materials

Slide Deck

Slides and session instructions for this module. Check the module status before teaching it.

Open presentation slides

Download editable slides

Session kit: prep, timing and rubric

Activity Worksheet

Learner worksheet aligned to the studio activity and rubric.

Open worksheet

Related Content

Datasets

• MouseConnects Dataset – Complete Hippocampal Connectome
• MouseConnects Connectomics Workflow

Background guides

• Julian, first-generation undergraduate
• Maya, graduate student

Tools

• Ask an Expert

Program guidance

• Research Incubator Model
• Education Models

Continue learning

• Career Pathways & Graduate School Prep