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 musty air of the storage unit pressed down on Emily. She was clearing out her mother's things. Her mother had been a whirlwind of energy, always on the go, always in control. Emily preferred the quiet. She found an old scrapbook, the pages yellowed and brittle. Inside, carefully tucked away, was a draft resignation letter from her mother's volunteer work.

The letter spoke of a feeling of being consumed, of not having any personal time, of the weight of responsibility that had come to feel… heavy. The last sentence echoed in Emily’s mind: "I don't know who I am without this." Emily’s breath caught in her throat. Her mother had always been Emily's compass.

Emily felt a sudden, unexpected wave of emotion. Her mother had always known what to do, how to be. Now, Emily felt a deep, unsettling feeling. She felt adrift, alone, and unsure of her own footing. She stared at the letter, its words a dark reflection of her own uncertain future. She needed her mother, her guidance, now more than ever.

Emotion: dependent

Cluster: Shame / Guilt
PC1 (Valence): -1.05 Negative
PC2 (Disposition): 0.69

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
de0.489
缺乏0.395
even0.310
either0.301
'0.297
Suppressed:
笑道-0.309
不久-0.293
一脸-0.263
затем-0.263
庆祝-0.258