01 Why Map the Brain

Technical Training: Nanoscale Connectomics

Session outcomes (60 minutes)

• Translate a broad neuroscience goal into one testable connectomics hypothesis.
• Define measurable structural outputs and one defensible null model.
• State one explicit non-claim to prevent over-interpretation.

Pedagogical arc

• Hook: why map structure at all?
• Model: question -> metric -> null -> boundary.
• Practice: learners draft and critique hypothesis briefs.
• Check: rubric-aligned exit ticket.

Visual opener: the motivation question

• Prompt: what specific scientific uncertainty is this figure trying to reduce?

Framing the evidence problem

• Structure is evidence of organization and constraints, not direct proof of dynamics.

Reverse-engineering analogy and its limit

• Good analogy for constraints.
• Bad analogy if used to claim full mechanistic causality from structure alone.

What structure can support (high-confidence)

• Motif enrichment/depletion hypotheses.
• Cell-type targeting bias quantification.
• Path-length and convergence/divergence constraints.
• Candidate priors for mechanistic or AI models.

What structure cannot prove alone

• Real-time state trajectories.
• Causal dynamics without perturbation/physiology.
• Full behavioral mechanism across contexts.

Workflow: question to claim

Biological question -> measurable endpoint -> dataset suitability -> null model -> interpretation boundary

Example A: recurrent microcircuit hypothesis

• Question: are triadic motifs enriched above local random expectation?
• Endpoint: motif counts normalized by degree/spatial constraints.
• Null: degree-preserving + distance-aware rewiring.
• Non-claim: enrichment does not prove online computation.

Example B: targeting specificity hypothesis

• Question: does class X preferentially target compartment Y?
• Endpoint: synapse-density ratio by compartment with uncertainty.
• Null: shuffled target labels preserving volume occupancy.
• Non-claim: specificity does not imply causal functional role.

Evidence quality gates before interpretation

• Reconstruction completeness threshold tied to claim type.
• Annotation agreement target for key labels.
• Error budget explicitly documented.
• Region/species/age boundary explicitly documented.

Common failure modes (teach explicitly)

• Claim inflation from descriptive results.
• Metric mismatch to hypothesis.
• Dataset scale mismatch to biological question.
• Post-hoc null-model selection.

Instructor discussion move (Think-Pair-Share, 6 min)

• Think: rewrite one overclaim into a bounded claim.
• Pair: identify missing metric/null information.
• Share: vote on strongest boundary statement.

In-class activity (12 min)

Draft one hypothesis brief containing:

1. question,
2. endpoint,
3. null model,
4. one confound,
5. one explicit non-claim.

Formative check rubric

• Pass: all five brief components present and coherent.
• Strong: endpoint and null align tightly to question.
• Flag: claims exceed available evidence class.

Exit ticket (3 min)

Write one sentence each:

• “Our data can support…”
• “Our data cannot support…”

Figure attribution and references

• Visual sources: internal outreach + module12 lesson assets (historical context).
• Pair with journal-club readings on structure-function boundaries.

External paper figure slots (add in final teaching run)

• Dense reconstruction figure (motif/cell-type evidence context).
• Multimodal connectomics figure (structure-to-function boundary context).
• Connectome-analysis methodology figure (null-model and inference limits).