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 world felt…flat. No peaks, no valleys, just an endless expanse of indifference. David found himself moving through his days as if in a fog, disconnected from everything and everyone. The news was a constant barrage of unpleasantness.

He ambled past a yard sale, the bright colours of the items a stark contrast to his internal landscape. His gaze snagged on an old teddy bear, sitting on a blanket, looking sad. It was familiar. He recognized the shape of the head, the missing eye, the stitching on the nose. It was his childhood bear, Barnaby.

He felt a pull in his chest, a sensation he hadn't experienced in weeks. He reached for Barnaby, his fingers tracing the worn fur. He paid, mumbled a thank you, and walked away, the bear clutched tightly to his side. The world didn’t look brighter, but something had shifted.

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