Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

The wind whipped around Ava, stinging her cheeks as she navigated the crowded yard sale. The whole experience felt like an intrusion, a grotesque display of other people's unwanted possessions. She’d come searching for vintage teacups, a small rebellion against the sterile minimalism of her new apartment, but the atmosphere left her feeling unsettled.

And then she saw him. Theodore, the bear. He was propped up against a rusty garden gnome, his once-bright fur faded and worn. She could almost remember the feel of it in her arms.

A wave of acrimony washed over her. She hated the way he sat there, a silent testament to a time when life felt uncomplicated, a life she could never return to. A life she didn't want.

“How much for that… item?” she asked, her voice flat, devoid of emotion. The man, the seller, pointed at a sign that said "Everything $1". She paid and left the bear.

Emotion: resentful

Cluster: Shame / Guilt
PC1 (Valence): -1.47 Negative
PC2 (Disposition): 1.50

Role in Research

This story is one of 1,000 stories generated for the emotion resentful. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 resentful stories, after denoising with neutral dialogue baselines, produces the resentful emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the resentful vector is projected through the unembedding matrix.

Promoted:
S0.485
aggravated0.263
愤0.261
👎0.245
L0.241
Suppressed:
own-0.318
soon-0.225
随后-0.225
serendip-0.219
🤩-0.210