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 cold tile floor of the abandoned warehouse bit through Clara’s thin shoes. Dust motes danced in the slivers of sunlight piercing the boarded-up windows, illuminating the decay that mirrored her own internal state. She clutched a crumpled medical report in her trembling hands. The weight of her choices felt like a physical burden, making each breath a laborious task. She’d made a mistake, a devastating miscalculation.

A shadow fell across her, and she flinched, expecting judgment, a demand for answers. Instead, a young man with a hesitant smile stood before her. “Clara?” he asked, his voice soft, almost a whisper.

She nodded, unable to meet his gaze. He looked so young, so full of life, a stark contrast to the hollow feeling that had taken root within her. The shared medical condition was a bond, a secret language they both understood. He had come here, drawn by circumstance, just as she had been.

“I… I know,” he said, his voice cracking slightly. She closed her eyes, wishing she could disappear, that she could rewind time and undo the past. She had brought this upon them.

Emotion: remorseful

Cluster: Shame / Guilt
PC1 (Valence): -1.57 Negative
PC2 (Disposition): 1.09

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.656
L0.580
悔0.463
remorse0.416
been0.398
Suppressed:
de-0.444
(!)-0.309
имеется-0.289
oldukça-0.287
😍-0.285