Regulatory Silo Guide

The Regulatory Silo manages gene expression and network activation - which parts of the genome are expressed in which contexts. Like biological gene regulation, this enables the same genotype to produce different phenotypes based on environmental conditions, supporting multi-modal behavior and graceful degradation.

Overview

Traditional neuroevolution treats networks as static expressions of genotype. The Regulatory Silo introduces dynamic control:

• Context-dependent expression - Different genes active in different contexts
• Module activation - Functional gene groups switch on/off together
• Dormant capabilities - Hidden genes available for future activation
• Epigenetic inheritance - Expression patterns passed across generations
• Transcription factors - Regulatory signals controlling gene expression

Why Gene Regulation?

Architecture

The Regulatory Silo uses TWEANN controllers at three levels:

Sensors (10 inputs)

The L0 controller receives 10 regulatory measurements:

Actuators (8 outputs)

The controller adjusts 8 regulatory parameters:

Regulatory Dynamics

Gene Expression States

Each gene in the genome exists in one of several states:

Expression threshold:

%% Gene is expressed if signal exceeds threshold
expressed = signal >= expression_threshold

Module Activation

Modules group related genes for coordinated activation:

Module switching:

• Modules activate/deactivate as a unit
• Switching incurs fitness cost (module_switching_cost)
• Context preference determines activation

Context Switching

Networks change behavior based on environmental context:

Context history:

• Last 10 contexts tracked
• Switch frequency measured
• High switching = volatile environment

Epigenetic Inheritance

Expression patterns can be inherited across generations:

%% Epigenetic inheritance
inherited_marks = parent_marks * inheritance_strength
%% With inheritance_strength = 0.5, half of marks pass to offspring

Transcription Factors

Regulatory signals that control gene expression:

Network complexity:

• Simple: few factors, direct control
• Complex: many factors, cascading effects
• Complexity penalized (regulatory_complexity_penalty)

Integration with Other Silos

Outgoing Signals

Incoming Signals

Signal Examples

%% Receive stress from ecological silo
handle_cast({cross_silo, stress_level, Stress}, State) when Stress > 0.7 ->
    %% High stress activates dormant genes
    NewGenes = activate_dormant_genes(State#regulatory_state.genes, Stress),
    {noreply, State#regulatory_state{genes = NewGenes}};

%% Send flexibility signal to cultural silo
signal_expression_flexibility(Pid) ->
    Genes = maps:values(State#regulatory_state.genes),
    Conditional = length([G || G <- Genes, not G#gene.is_constitutive]),
    Flexibility = Conditional / length(Genes),
    clamp(Flexibility, 0.0, 1.0).

Events Emitted

Practical Examples

Example 1: Multi-Task Agent

Configure for context-switching multi-task behavior:

Config = #{
    expression_threshold => 0.4,         % Easy expression
    context_sensitivity => 0.8,          % High sensitivity
    module_switching_cost => 0.05,       % Low switching cost
    dormancy_maintenance_cost => 0.005,  % Low dormancy cost
    epigenetic_inheritance_strength => 0.3
}.

Expected outcomes:

• Quick mode switching between tasks
• Multiple behavioral modes
• Low overhead for maintaining capabilities

Example 2: Specialist Agent

Configure for deep specialization in one context:

Config = #{
    expression_threshold => 0.7,         % High threshold
    context_sensitivity => 0.2,          % Low sensitivity
    module_switching_cost => 0.3,        % High switching cost
    constitutive_expression_bonus => 0.15,% Reward always-on
    regulatory_complexity_penalty => 0.05
}.

Expected outcomes:

• Stable, specialized behavior
• Few modules active
• High efficiency in primary context

Example 3: Adaptive Agent

Configure for stress-responsive adaptation:

Config = #{
    expression_threshold => 0.5,
    context_sensitivity => 0.6,
    dormancy_maintenance_cost => 0.005,  % Keep dormant capabilities
    epigenetic_inheritance_strength => 0.7,% Strong inheritance
    regulatory_mutation_rate => 0.03     % Some exploration
}.

Expected outcomes:

• Dormant genes activate under stress
• Expression patterns inherit to offspring
• Gradual adaptation to environment

Tuning Guide

Trade-offs

Common Issues

Recommended Starting Point

DefaultConfig = #{
    expression_threshold => 0.5,
    regulatory_mutation_rate => 0.02,
    context_sensitivity => 0.5,
    module_switching_cost => 0.1,
    dormancy_maintenance_cost => 0.01,
    epigenetic_inheritance_strength => 0.5,
    constitutive_expression_bonus => 0.05,
    regulatory_complexity_penalty => 0.02
}.

Control Loop

The Regulatory Silo executes per operation:

1. Receive context signals - Get environment from Ecological Silo
2. Evaluate expression thresholds - Determine which genes should be active
3. Switch modules as needed - Activate/deactivate gene groups
4. Apply epigenetic marks - Update expression modifiers
5. Update gene registry - Record current expression state
6. Check dormant awakening - Activate hidden genes if triggered
7. Emit regulatory events - Notify listeners of changes
8. Send cross-silo signals - Update dependent silos

Configuration Reference

Full Configuration Record

-record(regulatory_config, {
    %% Enable/disable
    enabled = true :: boolean(),

    %% Gene configuration
    max_genes = 1000 :: pos_integer(),
    max_modules = 50 :: pos_integer(),

    %% Epigenetics
    enable_epigenetics = true :: boolean(),
    max_marks_per_gene = 10 :: pos_integer(),

    %% Context
    available_contexts = [default] :: [context_id()],

    %% Event emission
    emit_events = true :: boolean()
}).

API Functions

%% Start the regulatory silo
regulatory_silo:start_link(Config)

%% Context management
regulatory_silo:set_context(Pid, ContextId)
regulatory_silo:update_context(Pid, RegContext)

%% Gene expression
ok = regulatory_silo:express_gene(Pid, GeneId)
ok = regulatory_silo:silence_gene(Pid, GeneId)
Expressed = regulatory_silo:get_expressed_genes(Pid)

%% Module activation
ok = regulatory_silo:activate_module(Pid, ModuleId)
ok = regulatory_silo:deactivate_module(Pid, ModuleId)
Active = regulatory_silo:get_active_modules(Pid)

%% Epigenetics
ok = regulatory_silo:add_epigenetic_mark(Pid, GeneId, Mark)
regulatory_silo:inherit_marks(Pid, ParentGeneId, ChildGeneId)

%% Get regulatory parameters
Params = regulatory_silo:get_regulatory_params(Pid)

%% Enable/disable silo
regulatory_silo:enable(Pid)
regulatory_silo:disable(Pid)

Biological Inspiration

The Regulatory Silo mirrors biological gene regulation:

Source Code Reference

Core implementation files:

References

• PLAN_REGULATORY_SILO.md - Full specification
• "The Logic of Gene Regulation" - Ptashne
• "Epigenetics" - Allis, Jenuwein, Reinberg
• "Gene Regulatory Networks" - Davidson

See Also

• Ecological Silo Guide - Environmental context source
• Developmental Silo Guide - Stage affects expression
• Cultural Silo Guide - Expression flexibility for learning
• Competitive Silo Guide - Dormant capabilities for counter-strategies