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 relentless fluorescent lights seemed to bore into him, exacerbating the inner darkness that had settled. He sat at his desk, staring out the window, watching the rain streak across the glass. The air felt heavy, and the sounds of the office grated on his nerves. He felt a profound sense of emptiness.

"Hey, Liam, I just booked for my trip this summer," said his colleague, Mark. "It's near Camp Starview."

Liam just shrugged. He was in no mood to deal with anyone’s holiday plans.

Mark continued. "Did you ever go to Camp Starview? It's the one with the telescope and the midnight hikes!"

Liam stiffened slightly. The name triggered a surge of memories, sudden and unexpected. He had gone there. "Yeah," he said, his voice barely audible.

"No way! That place was so much fun! I was in the Eagle cabin!" Mark then started describing his own time at the camp, and Liam let himself listen.

Emotion: gloomy

Cluster: Sadness / Despair
PC1 (Valence): -1.68 Negative
PC2 (Disposition): -1.37

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.368
无法0.319
😔0.300
いつも0.270
쓸0.255
Suppressed:
la-0.432
l-0.419
de-0.398
B-0.294
!-0.259