Skip to main content

Cortical Area Details Window

The Cortical Area Details window is the most comprehensive interface for viewing and editing cortical area properties. It provides access to all parameters that control how a cortical area behaves, processes information, and connects to other areas.

Opening the Details Window​

There are several ways to access this window:

From Circuit Builder:

  • Right-click a cortical area → Details
  • Double-click a cortical area node

From Brain Monitor:

  • Right-click a cortical volume → Details

From Quick Menu:

  • Select area(s) → Quick Menu → Details

Window Features​

The window consists of multiple collapsible sections, each controlling different aspects of cortical area behavior:

  1. Summary - Basic properties and dimensions
  2. Neuron Coding - Neural encoding parameters (IPU/OPU only)
  3. Neuron Firing Parameters - How neurons fire and accumulate charge
  4. Memory Parameters - Memory behavior (Memory areas only)
  5. Post-Synaptic Potential Parameters - Connection strength dynamics
  6. Cortical Monitoring - Visualization and data collection settings
  7. Connections - View and manage connections (single-select only)
  8. Advanced - Advanced properties and destructive operations

Each section can be collapsed or expanded by clicking the triangle button in its header.


Summary Section​

The Summary section contains the most commonly edited properties.

Basic Information​

Name

  • The friendly display name
  • Used in Circuit Builder and Brain Monitor
  • Can contain spaces and special characters
  • Editable for all cortical types

Region

  • Shows the parent brain circuit
  • Click to change which circuit contains this area
  • Affects organization in Circuit Builder hierarchy

Cortical ID

  • Unique identifier (read-only)
  • Generated by FEAGI
  • Used internally for all operations
  • Shown in Advanced section (Advanced Mode)

Cortical Type

  • IPU (Input), OPU (Output), Memory, Custom, or Core
  • Read-only after creation
  • Determines which sections are available

Dimensions​

Dimensions (X, Y, Z)

  • Size of the cortical volume in voxels
  • Larger dimensions = more neurons
  • Total neurons = X × Y × Z × neurons per voxel
  • Changes affect neuron count and NPU capacity

For IPU/OPU Areas:

  • Shows Dimensions Per Device
  • Total dimensions = dimensions per device × device count
  • Allows creating multi-camera or multi-sensor setups

Device Count (IPU/OPU only)

  • Number of input/output devices
  • Example: 2 cameras, 4 motors
  • Each device gets its own dimensional space

Position​

Position (X, Y, Z)

  • 3D coordinates in the genome space
  • Used for visual organization
  • Does not affect neural processing
  • Freely editable (even for Core areas)

Advanced Properties​

Neurons Per Voxel

  • How many neurons exist in each voxel
  • Default: 1
  • Higher values increase processing density
  • Hidden for IPU/OPU areas (they use different encoding)

Synaptic Attractivity

  • How attractive this area is for incoming connections
  • Range: typically 1-100
  • Higher values encourage more connections
  • Affects automatic synapse formation

Visualization Voxel Granularity

  • Controls rendering detail in Brain Monitor
  • Higher values = coarser visualization, better performance
  • Lower values = finer detail, more GPU load
  • Does not affect neural computation

Live 3D Preview​

When editing Position or Dimensions, a live preview appears in the Brain Monitor showing:

  • The cortical volume at its new location/size
  • Real-time updates as you adjust values
  • Visual feedback for spatial planning

Neuron Coding Section​

Available for: IPU and OPU areas only

Controls how sensory data is encoded into neural activity (IPU) or decoded from neural activity (OPU).

Data Type​

What kind of value the area carries, for example a percentage, a signed percentage, or a picture. It sets the shape of the area.

Absolute or Incremental​

  • Absolute - The reading or command right now
  • Incremental - A change since the last reading

Linear or Fractional​

How a percentage is laid out along Z:

  • Linear - One neuron fires, and its position along Z is the value
  • Fractional - Several Z neurons can fire, each worth half of the one before it

See Cortical Areas for how each is encoded.

Note: Changing coding parameters will update the Cortical ID automatically to reflect the new encoding scheme.


Neuron Firing Parameters Section​

Available for: Custom and Memory areas (not IPU/OPU)

Controls the fundamental behavior of individual neurons - how they accumulate charge, when they fire, and how they recover.

Membrane Potential Accumulation​

Toggle: ON/OFF

  • ON - Negative input is ignored, so inhibition cannot push the membrane potential below where it already is
  • OFF - Negative input lowers the membrane potential like any other input

Charge builds up over bursts the same way in both modes. The toggle only decides what happens to negative input.

Most areas use ON for temporal integration of signals.

Fire Threshold​

Range: Typically 0.0 - 1.0

The membrane potential level at which a neuron fires:

  • Lower values = neurons fire more easily
  • Higher values = neurons require stronger input
  • Critical for controlling area sensitivity

Threshold Limit​

Range: Integer value

The highest membrane potential at which a neuron can still fire:

  • A neuron fires only when threshold <= potential <= threshold limit
  • If the input pushes the potential above the limit, the spike is blocked and the neuron keeps its charge
  • The default (the largest value) means there is no upper bound

Neuron Excitability​

Range: 0-100 (percentage)

How easily neurons respond to input:

  • 100% = maximally responsive
  • 0% = completely inhibited
  • Useful for modulating entire area activity

Refractory Period​

Range: Typically 0-50 (bursts)

Time a neuron cannot fire after firing:

  • Prevents continuous firing
  • Creates temporal spacing between spikes
  • Measured in FEAGI burst cycles

Leak Coefficient​

Range: 0-100 (percentage)

How quickly neurons lose accumulated charge:

  • 0% = no leak, perfect integration
  • 100% = rapid decay, short-term only
  • Determines temporal integration window

Leak Variability​

Range: 0-100 (percentage)

Randomness in leak rate across neurons:

  • 0% = all neurons leak uniformly
  • 100% = maximum variation
  • Adds heterogeneity to temporal dynamics

Consecutive Fire Count​

Range: Integer value

Maximum times a neuron can fire consecutively before entering snooze:

  • Prevents sustained over-activation
  • Enforces temporal diversity
  • 0 = unlimited consecutive firing

Snooze Period​

Range: Integer value (bursts)

Rest time after reaching consecutive fire limit:

  • Longer than refractory period
  • Prevents neuron exhaustion
  • Encourages distributed activation

Fire Threshold Increment​

3D Vector: (x, y, z) values

How much the firing threshold grows with a neuron's position in the area:

  • A neuron's threshold is the base threshold + x * X increment + y * Y increment + z * Z increment
  • Neurons further along an axis are harder to fire
  • The threshold does not change as a neuron fires

Memory Parameters Section​

Available for: Memory cortical areas only

Controls how memories form, persist, and strengthen over time.

Initial Neuron Lifespan​

Range: Integer (bursts)

How long a new memory trace persists:

  • Short lifespan = working memory
  • Long lifespan = persistent memory
  • Starting point before growth mechanisms

Lifespan Growth Rate​

Range: Integer (bursts per activation)

How much lifespan increases when memory is accessed:

  • Higher values = faster memory consolidation
  • Zero = no strengthening (fixed lifespan)
  • Positive feedback mechanism

Long-term Memory Threshold​

Range: Integer (bursts)

Lifespan at which memory becomes permanent:

  • Memories above threshold never decay
  • Simulates long-term potentiation
  • Critical for persistent learning

Temporal Depth​

Range: Integer (bursts)

How far back in time the memory can recall:

  • Larger values = longer temporal window
  • Affects sequence learning capacity
  • Memory buffer size

Post-Synaptic Potential​

Available for: All cortical areas

Controls the strength and behavior of synaptic connections.

Post-Synaptic Potential (PSP)​

Range: Typically 0.0 - 1.0

Strength multiplier of this area's OUTGOING synapses (it is set on the source area):

  • A synapse delivers weight x PSP x sign (sign is -1 for inhibitory)
  • The value is stamped on each outgoing synapse when it is created; changing it later affects only synapses created afterwards (including learned ones)
  • If PSP Uniformity is OFF, the contribution is divided by the source neuron's number of outgoing synapses
  • Overridden by the membrane potential if MP-Driven PSP is ON
  • Plasticity never changes PSP; a synapse is silent if its weight or PSP is 0

PSP Max​

Range: Typically 0.0 - 2.0

Upper bound value stored in the genome and editable here:

  • The burst engine does not currently enforce it
  • It does not cap weights; the weight cap is the mapping's max_weight (set via API/MCP)

Degeneracy Coefficient​

Range: Typically 0.0 - 1.0

Amount subtracted from a source neuron's outgoing PSP each time it fires:

  • Per fire, not per time (floored at 0)
  • 0 = no degeneration
  • Does not change synapse weight

PSP Uniformity​

Toggle: ON/OFF

  • ON - Each synapse delivers its full weight x PSP x sign
  • OFF - That contribution is divided by the source neuron's number of outgoing synapses

MP-Driven PSP​

Toggle: ON/OFF

  • ON - The source neuron's membrane potential is used instead of the stored PSP
  • OFF - The stored PSP value is used

Cortical Monitoring Section​

Available in: Advanced Mode only

Controls visualization and data collection for analysis.

Membrane Potential Monitoring​

Toggle: ON/OFF Requires: InfluxDB connection

  • Logs membrane potential values over time
  • Enables time-series analysis
  • Stored in InfluxDB for historical data
  • Higher performance cost when enabled

Post-Synaptic Monitoring​

Toggle: ON/OFF Requires: InfluxDB connection

  • Logs synaptic activity over time
  • Tracks connection strength changes
  • Useful for plasticity analysis
  • Performance impact similar to membrane monitoring

Render Activity​

Toggle: ON/OFF

  • Controls visibility in Brain Monitor 3D view
  • ON: Show neural activity as colored highlights
  • OFF: Hide activity visualization
  • Does not affect neural computation
  • Improves performance for unused areas

Connections Section​

Available when: Single cortical area selected only

View and manage all neural connections to and from this cortical area.

Afferent Connections (Inputs)​

Lists all cortical areas that send signals TO this area:

For each connection:

  • Shows source cortical area name
  • Click name to open Mapping Editor window
  • View/edit connection parameters
  • Delete Icon Delete button to remove all mappings from that source

Add New Input:

  • Click + button in section header
  • Opens Mapping Editor to create new input connection

Efferent Connections (Outputs)​

Lists all cortical areas that receive signals FROM this area:

For each connection:

  • Shows destination cortical area name
  • Click name to open Mapping Editor window
  • View/edit connection parameters
  • Delete Icon Delete button to remove all mappings to that destination

Add New Output:

  • Click + button in section header
  • Opens Mapping Editor to create new output connection

Recursive Connections​

Shows if the area connects to itself:

  • "None Recursive" - No self-connection
  • "Recursive Connection" - Area connects to itself
  • Click button to create/edit recursive mapping

Use cases for recursive connections:

  • Pattern completion
  • Iterative refinement
  • Temporal integration
  • Attractor dynamics

Connection Navigation​

Click any connected area's name to:

  1. Open the Mapping Editor for that specific connection
  2. View connection topology and parameters
  3. Edit connectivity rule and pattern
  4. Adjust connection strength and plasticity

Advanced Section​

Available in: Advanced Mode only

Contains advanced properties and destructive operations that cannot be easily undone.

Delete Cortical Area​

Delete Icon Delete Button

Permanently removes the cortical area:

Confirmation dialog offers two options:

  1. Delete Area Only

    • Removes the cortical area
    • Preserves all connections (broken but stored)
    • Other areas remain intact
  2. Delete Area and All Connections

    • Removes the cortical area
    • Deletes all afferent connections (inputs)
    • Deletes all efferent connections (outputs)
    • Clean removal with no orphaned connections

Ctrl+Z (Cmd+Z on macOS) restores a deleted custom, memory, or input/output area and its mappings. The restored area gets a new id. Learned synapse weights are not restored. This lasts for the current genome session.

Note: Core areas have this button disabled - they cannot be deleted.

Reset Cortical Area​

Reset Icon Reset Button

Resets neural state to initial conditions:

What gets reset:

  • All neuron membrane potentials → 0
  • All neuron activation states → inactive
  • Accumulated charges and pending input cleared
  • Refractory and consecutive-fire counters cleared
  • Outgoing PSP restored to the area's configured value
  • Memory neurons (for Memory areas)

What is preserved:

  • All connections
  • All parameter values
  • Learned synaptic weights
  • Position and dimensions
  • Cortical area structure

Use cases:

  • Clear memory neurons and leftover activity
  • Start fresh training
  • Remove corrupted neural states
  • Debugging network behavior

Note: Core areas have this button disabled.

Additional Properties​

The Advanced section also displays:

Cortical ID (Read-only)

  • Unique identifier for API calls
  • Base64 encoded internal name

Unit Code (Read-only, IPU/OPU only)

  • Subtype identifier for device type

Subunit ID (Read-only, IPU/OPU only)

  • Individual unit index within device group

Multi-Selection Mode​

When multiple cortical areas are selected, the Details window adapts:

What Works in Multi-Select:​

Dimensions:

  • Edit dimensions for all selected areas at once
  • All areas get the same new dimensions
  • Useful for batch resizing

Parameters:

  • All firing parameters
  • All memory parameters
  • All PSP parameters
  • All monitoring settings

Batch Operations:

  • Changes apply to all selected areas
  • Conflicting values shown as "Multiple Selected"
  • Send button updates all areas simultaneously

What's Disabled in Multi-Select:​

Name: Cannot rename multiple areas at once Position: Cannot batch move (each area needs unique position) Connections Section: Only available for single-select Device Count: Cannot edit for multiple areas

Visual Indicators:​

"Multiple Selected" appears in fields when:

  • Selected areas have different values
  • Field cannot be batch-edited
  • No common value exists

Apply Button:

  • Stays disabled until changes are made
  • Clicking sends update to all selected areas
  • Shows progress/error for batch operations

Special Features​

Segmented Vision (isvi) Management​

When editing segmented vision areas (9-segment vision systems):

Automatic Layout:

  • Editing the center segment repositions all 8 peripheral segments
  • Editing any peripheral adjusts all 8 peripheral segments
  • Layout maintains proper spacing and alignment

Capacity Checking:

  • System validates NPU capacity before allowing resize
  • Warns if resize would exceed available neurons
  • Prevents changes that would crash FEAGI
  • Shows current/max neuron counts

Visual Feedback:

  • All 9 segments show live preview during editing
  • Warning color indicates capacity overflow
  • Apply button disabled if changes are unsafe

Core Area Restrictions​

Core areas (built-in FEAGI areas) have restrictions:

Read-Only:

  • Name (cannot rename)
  • Region assignment (cannot move)
  • Dimensions (cannot resize)
  • All neuron parameters
  • All memory parameters
  • All PSP parameters
  • Monitoring settings

Editable:

  • Position (can reposition for organization)
  • Connections (can modify mappings)

Disabled Operations:

  • Delete (Core areas are permanent)
  • Reset (Core areas maintain state)

This prevents accidental damage to essential FEAGI components.


Apply vs. Live Updates​

Understanding when changes take effect:

Immediate Updates (Live)​

These update instantly without clicking Apply:

  • Collapsing/expanding sections
  • Navigating between tabs
  • Opening connection editors
  • Changing focus between areas

Staged Updates (Apply Required)​

These require clicking the section's Apply button:

  • Any parameter changes
  • Dimension modifications
  • Position changes
  • Name edits
  • IO preset (Input, Output, or Interconnect)
  • Monitoring toggles

Apply buttons are disabled until changes are made

Section-Specific Apply Buttons​

Each section has its own Apply button:

  • Summary - Position, dimensions, name, IO preset, etc.
  • Neuron Coding - Coding parameters
  • Neuron Firing - Firing parameters
  • Memory - Memory parameters
  • PSP - Synaptic parameters
  • Monitoring - Visualization settings

This allows changing multiple sections and applying them independently.


Tips and Best Practices​

General Usage​

Start with Summary:

  • Get familiar with basic properties first
  • Understand the area's role before tuning parameters

Collapse Unused Sections:

  • Reduces visual clutter
  • Focuses attention on relevant controls
  • Window remembers your collapse states

Use Multi-Select Wisely:

  • Great for batch parameter tuning
  • Be careful - affects all selected areas
  • Verify selection before applying changes

Parameter Tuning​

Fire Threshold:

  • Lower threshold = more sensitive
  • Start around 0.5 and adjust
  • Watch activity in Brain Monitor

Leak Coefficient:

  • High leak = short-term memory
  • Low leak = temporal integration
  • Match to task requirements

PSP Values:

  • Start with uniform PSP
  • Enable learning only when needed
  • Monitor for runaway strengthening

Performance​

Disable Monitoring:

  • Turn off membrane/synaptic monitoring when not analyzing
  • Significant performance improvement
  • Only enable for specific debugging

Render Activity:

  • Disable for areas you're not watching
  • Reduces GPU load in Brain Monitor
  • No effect on neural computation

Visualization Voxel Granularity:

  • Increase for large areas
  • Reduce rendering load
  • Maintain detail only where needed