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).

"This isn't working," Clara whispered to herself, her voice barely audible over the hum of the server room. The data, the endless stream of code, felt like a suffocating blanket. Her head pounded. A sense of being lost, adrift in a sea of complexity, was weighing heavily on her.

She had been tasked with optimizing the company's new AI model, a task that once seemed exciting. Now, it had become a source of immense stress. Days bled into weeks, the lines of code blurring together until they formed a tangled, impenetrable mess. She'd stopped eating properly.

An email alert. Her project lead. A short, curt message: "Meeting - Algorithm Review." She closed her eyes. It felt like another shove towards a precipice.

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