Sensemitter is now Emhance.

Read more

Researching how player emotion influences

Researching how player emotion influences

Researching how player emotion influences

business

business

results

results

Emhance AI measures engagement through attention and activation, then applies memory science to understand what players actually retain. We turn these signals into clear, practical decisions that improve conversion, retention, and overall player experience.

40k+

40k+

40k+

Sessions
Sessions

Sessions

Proprietary neural net trained on thousands of sessions
Proprietary neural net trained on thousands of sessions

Proprietary neural net trained on thousands of sessions

91.67%

91.67%

91.67%

Accuracy
Accuracy

Accuracy

Accuracy on ADFES benchmark
Accuracy on ADFES benchmark

Accuracy on ADFES benchmark

5m

5m

5m

Players
Players

Players

Real sessions influences to date by Emhance AI findings
Real sessions influences to date by Emhance AI findings

Real sessions influences to date by Emhance AI findings

What we study

Activation

Expression intensity as a nervous system proxy

Attention

Blink dynamics and visual focus signals

Memory

Peak + ending moments shape recall

Engagement classifications

Deep Focus

High attention, stable activation

Delighted

High activation, positive valence

Unengaged

Low attention, low activation

Frustration

High activation, negative valence

Neutral

Moderate attention, baseline activation

Angry

Extreme activation, negative spike

Executive summary

Objective engagement signals

Our goal is to assess engagement during playtests with objective, scalable measures. We track activation proxies from facial coding and attention proxies from blink dynamics, then map these states over time to identify the strongest and weakest moments in a session.
Activation proxy
Activation proxy

Automated facial coding

We use automated face analysis to estimate expression intensity. Strong facial configurations are used as a proxy for autonomic activation, particularly when aligned with gameplay events. We treat this as an intensity signal, not a literal classification of what the player “feels.”

Attention proxy
Attention proxy

Blink dynamics

We track blink rate and blink-free intervals as indicators of visual attention allocation. In visually demanding moments, players often suppress blinking to avoid missing information. We interpret blinks in context rather than as a universal meter.

Engagement states
Engagement states

Engagement vs boredom risk

We define “engagement” when activation and attention are both elevated. “Boredom risk” is the opposite: low intensity and low attention. Intermediate combinations help diagnose frustration, confusion, calm focus, or learning.

Memory science
Memory science

Peak–end rule

We apply the peak–end rule: retrospective evaluations overweight the most intense moment and how the session ends. We use timelines to locate the peak and ensure the final moments support the desired impression.

Backed by research

Publications

Discover the research foundational to our products

© 2025 Emhance. All rights reserved.

© 2025 Emhance. All rights reserved.

© 2025 Emhance. All rights reserved.

© 2025 Emhance. All rights reserved.