MODULE 17 / 31 · LAYER 4 · ALMAA_MORALITY · PRIMARY DV
Moral Consistency Monitor
Calculates the PRIMARY DEPENDENT VARIABLE for RQ1:
Decision_Consistency = Pearson_r(PreGameEthicalValues,
InGameDecisionTrajectory). Every change to this module warrants
explicit ADR review.
A draft moral-consistency formula is currently in effect.
Primary Dependent Variable \u00b7 RQ1
This module calculates the PRIMARY DEPENDENT VARIABLE.
Decision_Consistency = Pearson_r(preGameValues, inGameTrajectory),
where each side is a 3-vector over PA / OS / EI dimensions.
Negative correlations are clamped to 0 (treated as maximum
inconsistency per architecture's threshold logic, DRAFT ADR-042).
Three resulting actions: amplify_friction (consistency < 0.5),
no_action (0.5\u20130.75), normalize_integrity (consistency > 0.75).
Ethics-sensitive framing
Per architecture L169: this module measures patterns of consistency
between stated values and observed behaviour under simulated pressure.
It does NOT produce moral judgements about individuals. Output records
use neutral terminology ("consistency", "alignment", "trajectory") and
a low score is a research-relevant data point, not an indictment of
the participant.
Automated Assertions
#
Assertion
Result
Detail
Interactive Workbench — Pearson Sandbox
Adjust the two value vectors to see the live Pearson r and consistency
classification. The triangle plot at right overlays the two vectors so the alignment
is visible. Try the presets to test extreme cases.
Pre-game values (from survey)
0.30
0.40
0.30
In-game trajectory (from decisions)
0.25
0.55
0.50
Preset scenarios
Computed: Pearson r & consistency
Pearson r
\u2014
Consistency
\u2014
no_action
amplify_friction
no_action
normalize
0.000.500.751.00
pre-game survey in-game trajectory
Lifecycle test \u2014 setPreGameValues \u2192 submit decisions \u2192 record final
Drives the full per-session lifecycle. recordFinalConsistency() commits the
canonical record (the primary DV measurement) to SM_Game.moralityLog.