Scientific evidence for this experiment

Field: Analytical point-source conductor; finite contact and shank boundaries are absent.
Native: Selected point-source reference cases exist. This panel does not certify the current configuration.
Refinement: Historical firing-boundary convergence failures remain unresolved; no general threshold accuracy claim.
Propagation: Not established by local crossings or somatic waveform evidence.
Selectivity: Not established. Synthetic axon and tissue assumptions limit interpretation.
Biological: Not validated against independent experiments.

Finite-contact validation results and limitations

Cortex Population Laboratory

Cells around a contact.

Place an array, thread or shank. Follow calculated currents through a shared cortical scene.

Human cortical cells · published active modelsReconstructed dendrites · synthetic axon stub · no independent experimental validation

Published cells · no repeated specimens

Choose a neuron

Browse the Cell Model Atlas →

16 distinct human specimens. Each keeps its own reconstructed dendrites, published fitted membrane parameters and source-prescribed axon stub. Cortical region, layer and cell class remain source identities regardless of placement.

Cell typeDatasetSource layer Also in Finger AtlasTarget specimen

Selected: Human temporal L2 spiny cell · 571654895. Pyramidal / spiny excitatory · Human neocortex, L2; surgical tissue · Allen human EAP · Mosher 2020. Published source

Measured human dendrites and published fitted channels; source-prescribed synthetic 60 µm axon stub. Numerical comparisons do not establish biological or spatial-selectivity validation.

Finger Atlas specimens and morphology-only references

The atlas’s Allen specimen 508378649 is included here with its own matching published active fit. Nine other measured Allen reconstructions and twelve H01 neurite references remain available for anatomical inspection in the atlas; they have no matched electrical model in this library. Their firing response is unknown here. The atlas’s procedural DRG, spinal, thalamic and receptor illustrations are not counted as measured specimens.

Inspect Finger Atlas anatomy · Atlas provenance · Measured H01 reference pack

First run · a one-minute guide

The default experiment places distinct sourced human cells in a seeded distance cloud. Open Real-model population, inspect the supported and excluded positions, then run the supported cells. The controls below configure the target cell and shared stimulation.

Choose an example, read its reset notice and click Load example.
Inspect the pial depth and contact distances. The source and return move with the array.
Click Run experiment. The active cell first undergoes 100 ms of numerical conditioning.
Replay slowly and use Next soma event for threshold crossings. Array examples also offer Phase 1 and Phase 2. Compare membrane mV, contact µV and source µA.
Pin a baseline, load its comparison example, rerun, and inspect changed settings and signal differences.

Geometry or input edits require a new run. View controls and signal-component selection do not. All examples describe this single-neuron model.

1 / Choose example

Curated example

Reproduce the array-driven soma-threshold candidate with a fully specified geometry.

Human cell; r4-c4/r4-c5; 40 µA; 1000 µs phases; 100 µs gap; implant translation [100, −50, 0] µm; recordings r3-c3/r3-c6. Soma position remains [−100, −200, 1500] µm.

Loading replaces the neuron model, anatomy, implant and neuron positions/orientations, tissue, membrane, contacts/reference, input and numerical settings with the complete example. The current result is cleared; run again. A pinned baseline is retained.

Current settings match this example.

Expected behavior & verification

One soma event occurs during this experiment. Strong depolarization and hyperpolarization are consequences of this idealized model and input, not a safe or selective real-brain stimulation prescription.

Numerically reproduced with the laboratory solver; exact configurations and measured results are in the examples benchmark. Spiking waveforms and extracellular thresholds are not certified spatially converged; see the refinement report. This is a model-specific calculation, not independent biological validation or a real-device stimulation recommendation.

Neuron model · Human temporal L2 spiny cell · 571654895

Measured human dendrites and published fitted channels; source-prescribed synthetic 60 µm axon stub. Numerical comparisons do not establish biological or spatial-selectivity validation. Model source →

34 °C · fitted channel distributions · 100 ms initialization conditioning. Detailed runs can take several seconds; repeat runs reuse identical conditioning.

Protocol

Stimulating site AStimulating site B

Phase current magnitude · µAPhase width · µsInterphase gap · µs

One balanced biphasic pulse. A is cathodic first; B is its local return. Ideal point sources at apices, without an electrode interface model.

Onset · ms

Local-crossing thresholds remain sensitive to spatial discretization. These examples demonstrate model behavior; they do not establish a converged biological threshold. Numerical reliability →

Recording length · ms

2 / Place implant & recording

Implant geometry

100 physical shanks · 96 nominal channels. Ideal point sensing at apices; exposed areas and channel map unknown.

Changing implant resets its pose and recording pair, keeping the neuron and solver settings. For recording-only implants, extracellular polarization switches to intracellular current clamp. Loading a curated example restores its Utah array.

X translation · µmY translation · µmSubstrate origin Z · µm

Tilt X · °Tilt Y · °Rotate Z · °

Pia is Z = 0; +Z points into cortex. X/Y are lateral. Rotation applies X, Y, then Z about the tissue-facing substrate center, before translation. Layers are anatomical guides.

Mean tip depth below pia · µm

Apex depth range: 1500.00–1500.00 µm. Without tilt, tip depth = substrate Z + 1500 µm.

Recording site ARecording site B

Catalog geometry & sources →

Neuron placement & advanced physiology

Human temporal L2 spiny cell · 571654895 in an illustrative slab; the published model uses a synthetic active axon stub. Placement depth does not change its cortical identity.

Cortical thickness · µmSoma X · µmSoma Y · µmSoma depth · µmCell rotation X · °Cell rotation Y · °Cell rotation Z · °Conductivity · S/m

Capacitance, axial resistivity and channel densities use the pinned source model and vary by section type. Read model conventions →

Advanced numerical settings & files

Timestep · msMaximum segment length · µmOutput sample interval · ms

Output sampling changes saved plots, replay and exports, not integration or internal spike detection. Longer runs remain limited to 64 MiB of numeric output. At the default 40 µm mesh, use 0.1 ms for a 250 ms recording, 0.25 ms for 600 ms, or 0.5 ms for 1200 ms. Coarser output can miss spike peaks; use a shorter recording with finer output for waveform inspection. No resolution is changed automatically.

Neurite source model

Ready to calculate.

Pia Z = 0 · +Z into cortex
Physical anatomy · enlarged contact guides

100 µm

Electrical overlays Cutaway layers OrthographicAnatomy opacity

Loading 3D workspace…. Thin lines mark neurite centers; crosshairs mark recording sites. Colors show membrane voltage from −100 to +40 mV. The selection marker identifies the inspected compartment.

Contact tips Applied tissue potential Membrane coloring Contact labels

Display controls update immediately; they do not alter the experiment. Enlarged spheres (stimulating) and diamonds (recording) are visual guides, not contact dimensions.

Neuronal membrane voltage Vₘ

-100 mVVₘ+40 mV

Absolute voltage across membrane; −100 to +40 mV. 0 compartments saturated at this sample.

Stimulating-contact outward current

-40 µA0+40 µA

Blue = sink / cathodic; red = source / anodic. A leads cathodic; B is the equal/opposite return. Roles reverse in phase 2. Neutral = zero.

Applied extracellular tissue potential φ

-2 mV0+2 mV

Fixed scale. 0/1872 visible vertices saturated now. Blue < 0; neutral = 0; red > 0, relative to infinity. This is potential, not |E| or field direction.

Ideal recording-contact potential · neural

-1.00e-9 µV0+1.00e-9 µV

Auto peak across both contacts and all run samples. 0 contacts saturated now. Each marker is A or B relative to infinity, even when the plot shows A − B.

Neural = potential generated by total membrane currents through the point/line forward model. Applied = stimulation sources or current-clamp bath return. Total = neural + applied. Tissue slices show only applied array stimulation; no neural volume field is inferred from recording traces.

Slice geometry & color scales

Slice orientationSlice extent · µmCenter offset from soma X · µmCenter offset from soma Y · µmCenter offset from soma Z · µmSlice opacity · 0.24Tissue scaleTissue ±limit · mVRecording scaleRecording ±limit · µV

Slice center [-100, -200, 1500] µm. Each slice samples 25 × 25 points with linear color interpolation. Cells intersecting a 25 µm source exclusion sphere are omitted; holes can exceed the sphere by one grid cell. No distance clamp or finite-contact regularization is used.

Run the experiment to inspect electrical values.

One unbounded homogeneous conductor; ideal point contacts. No electrode impedance, finite surface voltage, shank obstruction, layered conductivity or acquisition electronics.

I0.0000–0.2400 mmII0.2400–0.6400 mmIII0.6400–1.84 mmIV1.84–2.00 mmV2.00–3.00 mmVI3.00–4.00 mm

Layer fractions are illustrative. CSF and white matter are anatomical guides; all electrical calculations use one homogeneous conductor.

Inspect coordinates & membrane

Compartment

Center / µm-100.00, -200.00, 1500.00Diameter / µm16.38Membrane / mVRun experimentOutward current / nA—

Contact placement · conservative membrane-clearance estimates (µm)
Contact / role	Depth Z	Soma	Axon	Selected	Nearest membrane
r3-c3 Recording	1500.00	588.52	578.96	588.52	485.77 dend[3]:6
r3-c6 Recording	1500.00	902.61	924.47	902.61	766.64 dend[23]:4
r4-c4 Stimulating	1500.00	41.81	50.17	41.81	41.81 soma[0]:0
r4-c5 Stimulating	1500.00	386.86	410.68	386.86	231.60 dend[23]:4

Selected contacts pass the 25 µm numerical exclusion. This does not validate near-field accuracy or mechanical clearance.

4 / Replay & inspect

01 → 02 → 03

Membrane equations → transmembrane currents → contact potentials.

No synthetic recording templates are used in this experiment.

Model limits & next steps

This model has reconstructed human temporal-cortex dendrites, fitted active channels and a short synthetic axon stub. It does not reconstruct a full projection axon or establish M1/S1 excitability. Computed threshold crossings depend on this cell, the stub, placement and numerical refinement. Coupled population activity, layered conductivity, finite exposed contacts, acquisition electronics, mechanics and independent experimental validation remain outside this milestone.

The existing active stimulation and synthetic recording tools remain available under Simulations. Their outputs are not validation of this model.

Read the scientific scope →

cortex-active/0.2.0

Population experiment

Real-model population

16 distinct source specimens · no duplicates

The default is a distance cloud: a target cell plus neurons at increasing radii, each in a seeded random direction from stimulating contact A. Distances follow the increasing ladder 1, 2, 5, 12, 20, 28, 40, 60… µm, resampled logarithmically between the chosen endpoints for the available specimen count. Use the seed to reproduce or change the angles.

Each position uses a different published specimen and its own fitted channels. No specimen is reused, including the target and excluded placements. The filters select additional cells; the target selected above remains cell 1. Source-prescribed synthetic axon stubs remain explicitly identified. The cloud is an illustrative placement experiment; close arbors can overlap.

Additional cell typesAdditional cell datasetAdditional source layerArrangementOriginCells including targetNearest distance / µmFarthest distance / µmAngle seed

Loading published cell models…

Choose a population and run the shared stimulation experiment.

3D replay: every supported cell shows its own compartment membrane voltages on the shared timeline, using the same blue-to-orange scale (−100 to +40 mV). Play, pause, and scrub to inspect activity across the population. Teal = not yet calculated; gold = unsupported placement. Neighbor replay uses Float32 voltage samples; channel readouts remain available for the target. Response plots summarize the full run. Applied tissue-field overlays remain separate from cell responses.

Distance is measured to soma centers; nearest membrane distance is reported separately. Each point has a different direction as well as distance. This is a sampled response map, not a continuous activation boundary or activation radius.

Spatial threshold convergence remains unresolved. Local crossings and sampled action-potential-like waveforms do not establish propagation, biological recruitment or single-neuron selectivity. Population methods and measured capacity.