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).

He felt the knot in his chest tighten as the project manager, Mrs. Gable, reamed him out for a coding error. Leo felt an overwhelming need to retreat, to disappear. He craved the comfort of someone else’s guidance. His hands were trembling.

Suddenly, a voice, gentle yet firm, cut through his anxiety. It was Sarah, from accounting, offering him a tissue. "Long day, huh?"

He mumbled a weak agreement, wiping the sweat from his brow. He didn't want to be a bother. He was always worried about being a burden.

"I think I remember seeing you at Camp Skylark," Sarah said, a small smile playing on her lips. "I think you were in the archery class, right?" Leo's eyes widened. Camp Skylark. He thought of the archery instructor, a stern but ultimately caring woman who had taught him how to focus. He just looked at her, hoping.

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