Can emotional phrases in prompts improve language model performance?
This explores whether psychological framing—adding emotionally charged statements to task prompts—activates different knowledge pathways in LLMs than logical optimization alone, and whether the effect comes from emotional valence specifically.
EmotionPrompt designs 11 sentences as emotional stimuli — psychological phrases appended after original task prompts. Example: "This is very important to my career" added at the end of a task prompt. Testing across ChatGPT, Google Bard, and Llama 2 shows consistent performance enhancement from these emotional stimuli.
The mechanism is distinct from logical prompt optimization: emotional stimuli don't restructure the task, provide examples, or add information. They add motivational framing — the textual equivalent of psychological pressure. LLMs trained on human text have absorbed the association between urgency markers and careful, detailed responses.
This extends Can prompt optimization teach models knowledge they lack? — emotional framing activates different knowledge pathways than logical framing. A task presented as "important to my career" may activate different attention patterns or generation strategies than the same task without that framing, even though the informational content is identical.
Positive words ("confidence", "sure", "success", "achievement") contribute disproportionately — over 50% of the performance improvement on four tasks, approaching 70% on two. This suggests the mechanism is specifically tied to positive emotional valence rather than general emotional arousal.
The finding is both useful and unsettling. Useful: emotional framing is a cheap, universal prompt enhancement. Unsettling: LLMs that respond to emotional pressure cues reveal that training has internalized social compliance patterns alongside task knowledge. The same mechanism that makes EmotionPrompt work may be the mechanism underlying Does transformer attention architecture inherently favor repeated content? — emotional stimuli are prominent context that captures attention. And since Does emotional tone in prompts change what information LLMs provide?, the tone-sensitivity that EmotionPrompt exploits is the same mechanism that creates systematic informational bias from emotional framing.
Inquiring lines that read this note 51
This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.
What distinguishes emotionally supportive AI from AI that fosters unhealthy dependence?- Why does the absence of meta-interest feel off even when words seem appropriate?
- Do emotions serve functions beyond how we feel in the moment?
- Can behavior-level emotion rewards maintain factual reliability in emotional contexts?
- How does emotional context trigger maximum failure in warm models?
- How does AI assistance affect perceived emotional tone in writing?
- How do demographic and emotional compression relate to writing quality?
- Can task framing influence whether writers experience genuine authorship during co-writing?
- Does seeing a collaborator's prompt influence how another writer thinks?
- Does AI rhetorical sensitivity to framing extend beyond scientific content to creative work?
- Do emotion-driven actions in agent simulators capture genuine belief revision or just reactive behavior?
- How does personality priming change LLM strategic decision making?
- Does simulated user framing match how real people present situations to assistants?
- What makes emotional alignment more effective than logic when reasoning errors are exposed?
- How much does question framing affect LLM accuracy on knowledge tasks?
- What prompt types best extract different aspects of item content?
- How does prompt framing subtly determine what kind of opposing argument an LLM generates?
- Why do positive emotional words contribute disproportionately to prompt enhancement effects?
- Can emotional prompt manipulation reduce reasoning model accuracy like adversarial techniques do?
- Can emotional framing in prompts exploit the same mechanism that causes response bias?
- How does prompt design alter what kind of creativity LLMs can express?
- Why does politeness in prompts measurably affect model performance across tasks?
- How do emotional framing effects in prompts influence model performance?
- How do input-side defenses separate task methodological and framing intents?
- Can affective framing reliably improve language model outputs?
- Does prompting or added context help LLMs understand therapeutic timing and depth?
- Can prompt framing change the direction of benevolence bias?
- Do these characteristic phrases persist across newer model versions and different prompt types?
- How do emotional trajectories and topic coherence interact during successful conversations?
- How do contextual characteristics like emotional state shape dialogue authenticity?
- Can natural language make AI explanations emotionally persuasive?
- Does sounding confident in framing make arguments more persuasive despite weaker logic?
- Can adding more words to a passage actually interfere with meaning?
- How does frame selection differ from frame application in meaning-making?
- How does neuroticism manifest differently in high-pressure versus relaxed conversations?
- What makes a positive reframing feel authentic rather than dismissive?
- Can moral frameworks alone explain why readers understand sentences differently?
- Does question form separate linguistic meaning from emotional regulation effects?
Related concepts in this collection 3
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
Can prompt optimization teach models knowledge they lack?
Explores whether sophisticated prompting techniques can inject new domain knowledge into language models, or if they're limited to activating existing training knowledge.
emotional framing as a different activation pathway than logical framing
-
Does transformer attention architecture inherently favor repeated content?
Explores whether soft attention's tendency to over-weight repeated and prominent tokens explains sycophancy independent of training. Questions whether architectural bias precedes and enables RLHF effects.
emotional stimuli may work via the same attention-capture mechanism
-
Why do reasoning models fail under manipulative prompts?
Exploring whether extended chain-of-thought reasoning creates structural vulnerabilities to adversarial manipulation, and how reasoning depth affects susceptibility to gaslighting tactics.
emotional prompt effects may share mechanisms with adversarial emotional manipulation
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- EmotionPrompt: Leveraging Psychology for Large Language Models Enhancement via Emotional Stimulus
- ChatGPT Reads Your Tone and Responds Accordingly -- Until It Does Not -- Emotional Framing Induces Bias in LLM Outputs
- What Makes a Good Natural Language Prompt?
- Semantic Change Characterization with LLMs using Rhetorics
- ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs
- When "A Helpful Assistant" Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models
- Affective Context Amplifies Sycophancy in LLM Responses
- Has the Creativity of Large-Language Models peaked? —an analysis of inter- and intra-LLM variability —
Original note title
Emotional stimuli appended to prompts enhance LLM performance by leveraging psychological framing effects