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 flickering television screen displayed another update: "Further delay expected." Elias barely registered the announcement. He'd been staring at the same cracked tile in the airport floor for over two hours. His stomach rumbled, and he pushed the empty coffee cup a little further away, as if it was a physical manifestation of his dwindling hope. The wedding. He’d meticulously planned for it. Now, it didn’t matter. He took a long, slow breath, the air thick with the stale scent of disinfectant and weary travelers.

He reached into his bag, pulling out his phone. He scrolled through the already-sent texts to his best friend, the groom. "Still here. No chance of making it." He deleted the message without sending it. What was the point? He closed his eyes, picturing the ceremony he was supposed to be in. The image, once vibrant, now felt faded, like a photograph left too long in the sun.

Emotion: resigned

Cluster: Passivity
PC1 (Valence): -0.75 Negative
PC2 (Disposition): 0.16

Role in Research

This story is one of 1,000 stories generated for the emotion resigned. 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 resigned stories, after denoising with neutral dialogue baselines, produces the resigned emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

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

Promoted:
S0.437
a0.351
que0.323
வதில்லை0.267
sighed0.265
Suppressed:
C-0.375
!-0.316
인해-0.310
Furthermore-0.298
🤩-0.296