Proposed CDM revision — Modules 7–9 (Connectomics block)

Course: EN.585.781 Frontiers in Neuroengineering Module owner: Will Gray Roncal (WGR) Status: proposal for instructor and instructional-design review Companion material: rebuilt lecture decks at course/decks/marp/en585781/

Why this revision

Three things prompted it.

1. The instructional designer’s note on double-barrelled objectives. Melissa Rizzuto’s comment on the CDM — “need to choose one verb per learning outcome to avoid double-barrel objectives” — applies directly to modules 7–9. Six of the nine current MLOs carry two verbs or two objects.

2. The decks were rebuilt from current material. The field has moved: FlyWire’s whole adult fly brain (2024), the MICrONS flagship (2025), connectome-constrained models that predict activity (2024), and LICONN’s light-microscopy route to dense reconstruction (2025) all postdate the previous slides. The module topics should say so.

3. The three modules now share one spine. They walk a single discovery pipeline — question → specimen → image → reconstruction → graph → claim — with Module 7 owning the first and last columns conceptually, Module 8 the middle, and Module 9 the conversion of measurements into claims. The CDM should make that visible, because it is what makes the block cumulative rather than three topics.

Nothing here changes the assessment structure, the credit weighting, or the test schedule. Module 9 remains a test module (modules 7–9 assessed together).

Summary of changes

On dropping macroscale as a topic

Diffusion MRI and X-ray microtomography now appear once, as the example that different questions need different instruments — not as material to be covered. Two reasons: the course’s other modules already cover human non-invasive methods in depth (modules 10–12 in particular), and the connectomics block earns its place by teaching what only synapse-resolution structure can establish. Naming the boundary is pedagogically useful; re-teaching macroscale is duplicative.

Module 7 — Introduction to Connectomics

Proposed title: unchanged.

Proposed topics

• Why map the brain: the resolution and cost arguments, in numbers
• What structure can and cannot establish (the claim-bin framework)
• Scales, modalities, and representations — and the leakage between them
• The eight streams the field progresses along, and the results that mark them

Proposed MLOs (one verb each)

Change note. Current 7.1 (“Explain the scope … across scales”) and 7.2 (“Differentiate macro-, meso-, and nanoscale approaches”) overlapped. The replacement 7.2 is sharper and is the one students actually need: matching scale to question. New 7.3 makes the claim-sorting skill an explicit outcome — it is the block’s central competency and it was previously only implicit.

Instructional activities and assessments — unchanged in kind, plus: graded artifact — a one-page study brief containing a measurable endpoint, a null model, and an explicit non-claim.

Module 8 — Nanoscale Connectomics Tools and Methods

Proposed title: unchanged.

Proposed topics

• From tissue to voxels: preparation, sectioning, imaging, and their artifact signatures
• Storage and infrastructure: chunked multi-resolution arrays, serving, capacity and cost
• Reproducible, versioned pipelines: the ChunkedGraph, materializations, and provenance

Proposed MLOs

Change note. Current 8.2 (“Apply principles of reproducible pipelines”) and 8.3 (“Propose innovative strategies for scaling”) were both broad; 8.3 in particular had no observable artifact. The replacement 8.3 is assessable — a notebook that returns the same number a year later — and new 8.2 and 8.4 name skills the module already taught but never listed.

Instructional activities and assessments — unchanged in kind, plus: graded artifact — a reproducible query with a pinned materialization version, stated inclusion criteria, and one stated limitation.

Module 9 — Nanoscale Connectomics Algorithms and Applications

Proposed title: unchanged.

Proposed topics

• Segmentation, error taxonomy, quality metrics, and proofreading triage
• Graph construction and null models: the choices that determine the answer
• Applications, comparative connectomics, and NeuroAI

Proposed MLOs

Change note. Current 9.3 (“Synthesize and communicate algorithmic insights”) combined two verbs and two CLOs. Splitting it into 9.4 (justify) and 9.5 (assess) makes both gradable. Current 9.2 (“Interpret graph-theoretic models”) is subsumed by the sharper 9.3 and 9.4.

Instructional activities and assessments — unchanged in kind, plus: graded artifact — an analysis card (hypothesis, estimand, null model, success criterion, error band, non-claim, provenance). Test 3 covers modules 7–9 as scheduled.

Two smaller notes for the CDM as a whole

Module numbering and order. Brock Wester’s comments ask whether the modules are listed out of order and whether a preferred order should be re-established. From the connectomics block’s side: modules 7–9 are internally sequential and must stay in that order relative to each other, but the block as a whole can sit anywhere after the introductory modules. It does not depend on modules 1–6.

Module 16’s MLO numbering. Module 16’s learning objectives are currently numbered 14.1–14.3, duplicating module 14. Flagging it here since it is a one-line fix and this document is already going to the same reviewers. Not a connectomics-block issue.

What is already built

The three rebuilt decks — 59, 56, and 58 slides, each in three parts — are in the NeuroTrailblazers repository at course/decks/marp/en585781/, with rendered HTML and PowerPoint exports, and are published for community use at /teaching/lectures/ under CC BY-SA 4.0. They are written to the proposed MLOs above. If the MLOs change in review, the decks will be updated to match rather than the other way round.