• People are remarkably good at understanding
each other, and people can also predict others’
emotion and behavior.
• Emotion transition have some regularity.
• High probability: awe (畏敬) → gratitude (感謝)
• Low probability: awe (畏敬) → contempt (軽蔑)
• Then, people learned these rules based on own
experiences, and might be able to gain accurate
mental models of others’ emotion transition.
Studies in this article
• Study 1-3: the actual rates of transitions between
emotions using existing experience-sampling
datasets (own experienced and mental models for
• Study 4: Markov modeling over a rich sampling
over a rich sampling of 60 states + conceptual
building blocks of mental models.
• Study 5: Tyranny of the majority (数の暴力: 200万)
+ emphasis on dynamics
There is the strong associations
NRMSE = normalized root mean square
error in simple regressions with ratings
Conceptual building: Study 4
• Each dimension evaluated in previous studies.
• These index explained highly similar transition in
specific valence (e.g., negative: angry ⇒ sad)
• The above relationships for each dimensional
evaluation is Static information and association.
⇒ controlling for the four conceptual dimensions.
⇒ unique knowledge about emotional dynamics:
residual accuracy (mean partial ρ =0.10) + statistical
significant [95% bootstrap CI = (0.09, 0.11)], with a
large standardized effect size (d = 1.51).
• Almost results were replicated using the 2
million data from the Experience Project
The current results
• The mental model seems to be common as
own emotional transition.
• The correlation between own emotional
transition and mental model remains
meaningful with controlling static structure
using some dimensional evaluations.
• Participants in this study did not make
predictions in a naturalistic context.
• We used mental model for other persons based
on our emotional transitions.
That is, our emotion predicted others’ emotion!