Featured Research

From Sentiment
to Sentient

Every NLP tool reads tone. Sentient analysis models opposing parties as cognitive agents, with belief architectures, wound stacks, and behavioral modifiers, then predicts what they'll do next.

Colin McNamara · 2026

The Distinction

Two Approaches, One Word Apart

Sentiment and sentient share nine letters. They share almost nothing else.

Sentiment Analysis

Reads document tone
Flags positive / negative / neutral
Backward-looking: what was said
Surface-level NLP

Sentient Analysis

Models cognitive architecture
Maps 5-tier behavioral modifiers
Forward-looking: what they'll do
Agent-level prediction

The gap between sentiment and sentient is the difference between reading a transcript and understanding a mind.

The Architecture

The Five Tiers

Sentient analysis structures its cognitive model across five tiers, each one deeper into the decision-making architecture of the opposing party.

Belief Architecture

Core identity beliefs, self-narratives, and non-negotiable positions. The bedrock layer that shapes how all incoming information is filtered and interpreted.

Motivation Stack

What drives action: financial survival, ego protection, revenge, justice. The ordered hierarchy of motivations that determines which outcomes a party will pursue, and which they'll sacrifice.

Emotional State

Current activation level, threat response mode, and decision-making capacity. A party operating in amygdala hijack makes fundamentally different choices than one in executive function.

Cognitive Impairment

Stress degradation, confirmation bias amplification, and tunnel vision. Under sustained litigation pressure, cognitive function narrows. The narrowing is predictable.

Coordination Dynamics

Advisor influence, group polarization, and information asymmetry. No party acts alone. The orbit of attorneys, advisors, and allies shapes every decision through distortion or clarity.

The Evidence

Validated, Not Theoretical

Predictions are logged before outcomes occur. No post-hoc fitting. No cherry-picking.

91%

Acquit Score™ Calibration

201

Tracked Predictions

Behavioral Tiers

Every prediction is logged before the outcome occurs. No post-hoc fitting. No cherry-picking. The framework is tested against real-world litigation dynamics where the stakes are maximum and the actors are under extreme stress.

The Foundation

Sentient Analysis Requires Pristine Evidence

You can't model a mind from corrupted inputs. Sentient analysis depends on the same six-capability evidence pipeline that makes Acquit.ai's intelligence defensible, and it feeds directly into the ARPN Framework, where the five behavioral tiers above power the Behavioral Amplifier (BA) dimension. FMEA scores the failure. Sentient analysis explains why the failure will happen.

Collect

Verify

Classify

Brief

Present

Prove

See It In Action

Live demonstration of sentient analysis on anonymized case data.

Get Your Acquit Score

Presenting “From Sentiment to Sentient” at the AI & Law Summit, March 27

Apply the methodology: Red Team Your Case (ARPN Framework) → Deep dive into the Behavioral Modifiers Framework →

Join the waitlist for early access to the Acquit Score platform.

Sentient Analysis Methodology © 2026 Colin McNamara / Acquit.ai. Licensed under Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). Attribution required for academic and professional use. Commercial licensing: colin@acquit.ai