fly-api · plain-language explainer

How the Fly Learns

The demo's fly brain learns that one smell means a treat is coming. But the model is a wiring diagram — a map of a brain, not a brain. So what actually changes when it learns? A small set of connections gets weaker, by the rule real flies use. Five pictures tell the story; the numbered notes hold the detail.

On a phone, tap a note number to open it.

1 · The fly's memory organ

smell sensor cells 2,279 in the antennae (ORNs) sorting station 1,114 cells (antennal lobe) memory cells 5,177 — each smell wakes only ~1–4% (Kenyon cells) APL (2 cells) a damper that keeps most memory cells quiet the only connections that change decision cells 96 (MBONs) approach / avoid something good reward cells (323) dopamine neurons: PAM reward · PPL1 punishment "change these connections now"
The smell-to-memory circuit, from the fly's actual wiring diagram, with the number of cells of each kind in our model. Amber: the only connections that change. Red: the cells that order the change.

A smell wakes a few dozen of the five thousand memory cells — a different few dozen for each smell.The memory cells are Kenyon cells (KCs). Each listens to a random handful of the 685 projection neurons leaving the antennal lobe and fires only when several agree; the two giant APL neurons feed inhibition back so only the best-matched cells win (sparse coding). In our runs each odor activates 1–4% of KCs, two odors' codes overlap by ~1% (Jaccard 0.01), and the same odor gives the same code every time (Jaccard 1.00). Each memory cell connects to a few of about a hundred decision cells, which vote approach or avoid; reward cells fire when something good happens.Decision cells = mushroom-body output neurons (MBONs), 96 in 35 types. Reward cells = dopamine neurons: 307 PAM (reward) and 16 PPL1 (punishment). Counts come from the model's cell-type labels (experiments/learning/mb_subnet_ids.json). Those memory-to-decision connections are the knobs. Reward cells are the hand that turns them.

2 · How do we know this is where flies learn?

1 · break it memory organ no smell memory (everything else works) Heisenberg 2003 2 · fake the reward smell light ⚡ reward cells no sugar at all — the fly learns the smell anyway Claridge-Chang 2009 · Aso 2014 3 · listen to the knob before training after training memory → decision connection gets weaker, not stronger Hige 2015 4 · the trigger dopamine receptors on the connection detect "smell + reward together" Cohn 2015 · Handler 2019
Four experiments, twenty years. Each was done in living flies with cell-type-specific genetics; together they fix the place, the trigger, and the direction of the change.

For most brains, including ours, nobody knows the mechanism of learning at this level. The fly's memory organ is the exception. Damage it and flies can't learn smells.Heisenberg 2003 reviews the classical mutant and lesion evidence. Aso et al. 2014a showed the anatomy reads as an architecture for learning: a sparse KC code feeding 15 compartments, each pairing specific dopamine neurons with specific MBONs. Flash the reward cells with light instead of giving sugar, and flies learn anyway.Claridge-Chang et al. 2009 ("Writing memories with light-addressable reinforcement circuitry"); Aso et al. 2014b mapped which dopamine neurons write reward vs. punishment. This optogenetic substitution is why our simulation can deliver reward directly at the dopamine neurons (§5). Record a memory-to-decision connection before and after training and it is weaker, not stronger.Hige et al. 2015, by patch-clamp: pairing an odor with dopamine-neuron activity depresses that odor's KC→MBON synapses in the matching compartment. And dopamine, released by the reward cells onto exactly those connections, is the trigger.Cohn, Morantte & Ruta 2015: dopamine release is compartmentalized and state-dependent. Handler et al. 2019: the sign and timing of the plasticity follow from two dopamine receptors with different kinetics. Modi, Shuai & Turner 2020 review the whole story; Bennett et al. 2021 is one of several computational models built on this rule.

3 · The map

microscope one fly brain, sliced into thousands of layers and imaged at nanometre scale wiring map 139,255 neurons traced, ~50 million connections, each with a synapse count FlyWire · Dorkenwald 2024 labels every neuron named: sensor, memory, decision, reward cell … Schlegel 2024 simulation each neuron: fill up, fire, reset each connection: as strong as its synapse count Shiu 2024 not in the map how connections change with experience (the learning rule) · receptor types · dopamine release · gap junctions added by hand
From microscope to simulation. The map records where every connection goes and how many synapses it has. It records nothing about how connections change.

In 2024 one team traced every neuron and connection in a fly brain from microscope images,FlyWire (Dorkenwald et al. 2024): 139,255 neurons and ~50M synapses from one adult female, community-proofread, with a synapse count for every connected pair. Schlegel et al. 2024 add a class and type for every neuron — that is how we know which cells are sensor, memory, decision, or reward cells; the groups in the diagrams are selected by those published labels, not by us. and another turned the map into a simulation: each neuron a simple fill-up-and-fire unit, each connection as strong as its synapse count.Shiu et al. 2024: leaky integrate-and-fire neurons with published constants we did not change — rest/reset −52 mV, threshold −45 mV, membrane time constant 20 ms, synaptic decay 5 ms, refractory 2.2 ms, delay 1.8 ms; each synapse contributes 0.275 mV, signed by the predicted neurotransmitter. The model reproduces known responses (sugar → proboscis extension) with no fitting. A map shows where the roads go, not how traffic reshapes them: it contains no learning rule. Any such rule is added by hand, so the honest question is whether it is the right rule in the right place.Also absent: receptor identities, neuromodulator release, gap junctions, short-term synaptic dynamics. For this demo we cut the map to the olfactory pathway plus mushroom body — 8,991 neurons, 792k synapses — which surfaced a finding: the published whole-brain model is bistable, and almost any central input tips it into a permanent ~8,400-neuron storm. Four structural edits (each biologically argued, none a parameter fit) remove it while leaving odor coding intact; listed in the appendix, audited in notes.md.

4 · The rule

before memory cells decision cells all knobs at full strength smell A + reward A's crew fires reward cells "cut these in half" only the active cells' knobs are touched after A's grip on the decision is halved; every other smell is untouched
One learning step. Smell A's crew of memory cells fires while the reward cells fire; the connections leaving that crew are halved. Everything else is left alone.

The model's memory is a list of about 62,000 numbers — the strength of every memory-to-decision connection — and nothing else carries over from one moment to the next.62,261 KC→MBON connections (recomputed from the subnet's cell lists and the FlyWire connectivity table). Weights start at the connectome value; membrane voltages and synaptic conductances are reset between episodes, so this vector is the only persistent state.

Whenever the reward cells fire, every connection coming out of a memory cell that was active at the same time gets cut in half.

That is the whole algorithm.Verbatim from learning_driver_mb.py:kc_active = KCs with ≥1 spike this episode pam_hz = mean PAM DAN rate this episode if us and pam_hz >= 1.0: # dopamine gate hot = synapses whose presynaptic KC is in kc_active w[hot] *= (1 - eta) # eta = 0.5Dopamine-gated long-term depression at KC→MBON, once per episode. No objective, no gradient, no hidden state. Its place, trigger, and direction are the fly's (§2). The fine print is ours: one step at the end of each half-second instead of a gradual, timing-sensitive change;Real plasticity is continuous and receptor-mediated, with sign and size depending on odor–dopamine timing at sub-second resolution (Handler 2019). Our "same episode = coincident" abstraction cannot reproduce forward-vs-backward pairing effects. one reward signal for all decision cells instead of many;Biology: 15 compartments, each dopamine-neuron type writing to its own MBON zone, appetitive and aversive traces in parallel (Aso 2014, Cohn 2015). Model: one scalar gate (mean PAM rate ≥ 1 Hz) and one uniform η. Real dopamine neurons also make fast synapses; we zeroed those for stability, so dopamine acts only through the scripted rule. and memories that never fade.Depression only, permanent: no recovery, extinction, or forgetting. The full biology-vs-model ledger is in the appendix.

5 · Teaching the fly

1 · present the smell half a second of brain activity switch on smell A's sensor cells on reward trials, also switch on the reward cells 2 · take notes which memory cells fired for this smell? 3 · was there a reward? did the reward cells fire during this episode? no → nothing changes 4 · turn the knobs down halve every connection from those memory cells to the decision cells next episode: everything resets except the connection strengths — they are the memory
One training episode, repeated five times for smell A with reward, interleaved with smell B without. The knob-turning is a bookkeeping step between episodes, not part of the simulated brain.

Pavlov's dog drooled at the bell; a fly walks toward a smell that came with sugar.The smell is the conditioned stimulus (CS), the sugar the unconditioned stimulus (US). The classic assay is Tully & Quinn 1985: train with odor A + sugar or shock and odor B alone, then let flies choose between A and B in a T-maze; the fraction choosing A is the memory score. B is the control — it experiences everything except the smell–reward coincidence. We present smell A with the reward cells switched on, five times, with smell B and no reward in between.CS: a disjoint set of six ORN receptor classes, count-balanced, driven at 500 Hz for 0.5 s. US: the 307 PAM dopamine neurons at 60 Hz. We inject the US at the dopamine neurons because the model's own sugar→PAM route is silent in this regime (measured: PAMs at 0 Hz under sugar drive) — a documented limitation with a real experimental precedent (§2, experiment 2). Afterwards the decision cells' response to A has dropped by about 99%; B is untouched.Five pairings at η = 0.5 compound to ~97% depression per trained synapse. A's MBON response falls from 3,700–4,300 spikes to 0–57 across three seeds; B stays at 3,400–4,300. The spike readout uses a ×20 efficacy correction on KC→MBON synapses; the synaptic-trace readout needs none and shows the same picture. We cannot watch this fly walk to a T-maze arm here (the navigation demo does that), so the decision cells' response stands in for behavior. Sensitivity in the report. Smells that partly resemble A are partly suppressed — that is the wiring's doing, not the rule's.Probe odors sharing 67/50/33/0% of A's receptor classes inherit graded suppression (~60–75% at 67% overlap, falling to 0 at disjoint), because they share memory cells and therefore share the turned-down knobs. The synaptic-trace version of this gradient needs no tuned parameters at all. It is the qualitative signature of generalization gradients in real flies.

So: realistic, or a hack? Both, cleanly separated. The place, trigger, and direction of the change are the fly's (§2); the way similar smells share a memory comes from the real wiring (§3). Ours is the fine print: one big step, one reward signal, memories that never fade. The test of "realistic" is whether the result depends on our choices; the one place it partly does — how we read out the decision cells — is flagged in the notes, and the similarity result survives with nothing tuned at all.

Appendix · the assumptions ledger

For the technically inclined: every place the model is a cartoon of the biology, and why it matters.

AspectReal flyThis modelConsequence
Synaptic weight Conductances, receptor content, short-term dynamics Synapse count × predicted transmitter sign (Shiu 2024) Crude proxy; underlies the readout-gain caveat below
Substrate stability Stable via neuromodulation, gap junctions, adaptation — none in the wiring diagram Four structural edits: DAN fast outputs → 0; KC→KC → 0; ORN afferents input-only; excitatory AL local-neuron outputs → 0 The substrate needed surgery before learning was testable — a finding about connectome models, not flies
Dopamine's fast action DANs also make fast synapses (co-transmission) Zeroed (edit above); dopamine acts only through the scripted rule Clean separation, but neuromodulation exists only where we wrote it in
Update dynamics Continuous, receptor-mediated; sign and size depend on odor–dopamine timing at sub-second resolution (Handler 2019) One batch multiply between 0.5 s episodes; coincidence = "same episode" Timing-dependence (forward vs. backward pairing) cannot emerge
Dopamine signal Compartmentalized: 15 compartments, each DAN type writing to its own MBON zone, opposite valences in parallel (Aso 2014, Cohn 2015) One scalar gate — mean PAM rate ≥ 1 Hz — and one uniform η for all KC→MBON synapses No appetitive/aversive asymmetry, no parallel memory traces
Direction of change Bidirectional: depression and recovery; memories decay and extinguish Depression only, permanent Extinction, forgetting, re-learning out of scope
Reward pathway Sugar sensing recruits PAM DANs through circuitry PAMs driven directly at 60 Hz — the connectome's own sugar→PAM route is silent in this LIF regime Mirrors optogenetic conditioning, a real paradigm — but the natural reward path is not demonstrated
Behavioral readout MBON compartment balance → downstream motor bias MBON spike counts, with a ×20 efficacy correction on KC→MBON synapses; the synaptic-trace readout needs no correction Headline % depends on readout choice; the parameter-free trace shows the same gradient

References

  1. Tully & Quinn (1985). Classical conditioning and retention in normal and mutant Drosophila melanogaster. J Comp Physiol A. doi:10.1007/BF01350033
  2. Heisenberg (2003). Mushroom body memoir: from maps to models. Nat Rev Neurosci. doi:10.1038/nrn1074
  3. Claridge-Chang et al. (2009). Writing memories with light-addressable reinforcement circuitry. Cell. doi:10.1016/j.cell.2009.08.034
  4. Aso et al. (2014a). The neuronal architecture of the mushroom body provides a logic for associative learning. eLife. doi:10.7554/eLife.04577
  5. Aso et al. (2014b). Mushroom body output neurons encode valence and guide memory-based action selection. eLife. doi:10.7554/eLife.04580
  6. Cohn, Morantte & Ruta (2015). Coordinated and compartmentalized neuromodulation shapes sensory processing. Cell. doi:10.1016/j.cell.2015.11.019
  7. Hige et al. (2015). Heterosynaptic plasticity underlies aversive olfactory learning in Drosophila. Neuron. doi:10.1016/j.neuron.2015.11.003
  8. Handler et al. (2019). Distinct dopamine receptor pathways underlie the temporal sensitivity of associative learning. Cell. doi:10.1016/j.cell.2019.05.040
  9. Modi, Shuai & Turner (2020). The Drosophila mushroom body: from architecture to algorithm. Annu Rev Neurosci. doi:10.1146/annurev-neuro-062119-092917
  10. Bennett, Philippides & Nowotny (2021). Learning with reinforcement prediction errors in a model of the Drosophila mushroom body. Nat Commun. doi:10.1038/s41467-021-22592-4
  11. Dorkenwald et al. (2024). Neuronal wiring diagram of an adult brain. Nature. doi:10.1038/s41586-024-07558-y
  12. Schlegel et al. (2024). Whole-brain annotation and multi-connectome cell typing of Drosophila. Nature. doi:10.1038/s41586-024-07686-5
  13. Shiu et al. (2024). A Drosophila computational brain model reveals sensorimotor processing. Nature. doi:10.1038/s41586-024-07763-9

Back to the demo · full method and audit trail in the learning report.